The Transfer Pricing Officer has rejected our margin and adopted the median of his comparables. When is he entitled to go to the median at all, how is the range built, and what happens if he is left with only four or five comparables?
The range is built only where the dataset has six or more entries and the most appropriate method is not the profit split method or the other method. Rule 10CA(2) requires a dataset constructed by placing the prices in ascending order. Rule 10CA(4) then provides that where the most appropriate method applied is a method other than the method referred to in clause (d) or clause (f) of s.92C(1) and the dataset consists of six or more entries, "an arm's length range beginning from the thirty-fifth percentile of the dataset and ending on the sixty-fifth percentile of the dataset shall be constructed". Rule 10CA(5): if the price actually charged is within that range it "shall be deemed to be the arm's length price" — no adjustment. Rule 10CA(6): if it is outside the range, "the arm's length price shall be taken to be the median of the dataset". The median is therefore reached only on the footing that the range was properly constructed and the price fell outside it; there is no route to the median in a case where sub-rule (4) never applied. Where sub-rule (4) does not apply — fewer than six entries, or the profit split method or the other method — sub-rule (7) revives the older machinery: the arm's length price is the arithmetical mean of all the values in the dataset, subject to the proviso that if the variation does not exceed such percentage not exceeding three per cent as may be notified, the price actually charged is deemed to be the arm's length price. The thirty-fifth percentile, the sixty-fifth percentile and the median are all defined in sub-rule (8) on the same pattern, and the pattern is not the ordinary statistical one: each is the LOWEST value in the ascending dataset such that at least thirty-five, sixty-five or fifty per cent of the values are equal to or less than that value, with a proviso that if the number of values equal to or less than that value is a whole number, the percentile is the arithmetic mean of that value and the value immediately succeeding it.
Decided by the CBDT Circulars & Instructions (Not applicable — statutory text) on 2015-10-19, reported as Rule 10CA of the Income-tax Rules, 1962, heading "Computation of arm's length price in certain cases", transcribed from https://www.incometaxindia.gov.in/w/rule-10ca (no "Year:" stamp; the page prints "Upload Date: 13/12/2025") and corroborated on https://incometaxindia.gov.in/Rules/Income-Tax%20Rules/ITRule10CA.htm (no date stamp of any kind); rule 10B read on https://www.incometaxindia.gov.in/w/rule-10b for the methods referred to in sub-rules (2) and (3) of rule 10CA, and s.92C(1) clauses (d) and (f) read on https://www.incometaxindia.gov.in/w/section-92c-24 (Year: 2025). It bears on section Rule 10CA, section Rule 10CA(2), section Rule 10CA(3), section Rule 10B, section Rule 10B(2), section Rule 10B(3), section Rule 10B(4), section Rule 10C(2), section 92C, section 92C(1), section 92C(2), section 92CA, section 92CA(3), section 92B, section 92BA of the Income Tax Act 1961, in Assessment & Scrutiny, How Tax Law Is Read and Evidence & Burden of Proof matters.
Three things go wrong on these facts and all three are worth taking. First, the six-entry threshold is jurisdictional to the range. An officer who has five comparables left after exclusions cannot construct a range and cannot adopt the median: he is in sub-rule (7) and must take the arithmetical mean, with the notified variation available on top of it. That is a materially different number and it is the single most common error on a small comparables set. Second, sub-rule (4) is disapplied not only by the size of the dataset but by the method — the profit split method in s.92C(1)(d) and the other method in s.92C(1)(f) are excluded from the range machinery altogether. Third, the sub-rule (8) definitions do not produce the percentile an ordinary spreadsheet function produces. The rule works by counting values that are equal to or less than a candidate value and taking the lowest value at which the count reaches the threshold; the proviso then averages that value with its immediate successor, but only where the COUNT of values equal to or less than it is a whole number. A percentile computed by interpolation, which is what most software does, will not match, and a margin that the officer's spreadsheet puts outside the range may be inside it on the words of the rule. The library already holds Barracuda Networks on the use of multiple year data in the same rule; this entry supplies the text that decision is applied to.
Binding on the department, not on the assessee or the courts. An assessee may rely on a circular that is beneficial to them.
Read aloud by your device. Press again to stop.
Rule 10CA, as printed on the departmental page, reads in material part: "(1) Where in respect of an international transaction or a specified domestic transaction, the application of the most appropriate method referred to in sub-section (1) of section 92C results in determination of more than one price, then the arm's length price in respect of such international transaction or specified domestic transaction shall be computed in accordance with the provisions of this rule. (2) A dataset shall be constructed by placing the prices referred to in sub-rule (1) in an ascending order and the arm's length price shall be determined on the basis of the dataset so constructed" — followed by three provisos. The first governs "a case referred to in clause (i) of sub-rule (5) of rule 10B, where the comparable uncontrolled transaction has been identified on the basis of data relating to the current year" and the comparable enterprise has in either or both of the two financial years immediately preceding the current year undertaken the same or similar comparable uncontrolled transaction, in which case the weighted average of the prices computed in the manner provided in sub-rule (3) is included in the dataset instead of the price referred to in sub-rule (1). The second governs "a case referred to in clause (ii) of sub-rule (5) of rule 10B, where the comparable uncontrolled transaction has been identified on the basis of the data relating to the financial year immediately preceding the current year" and the comparable enterprise has in the financial year immediately preceding those two financial years undertaken the same or similar comparable uncontrolled transaction, in which case the weighted average of the prices of the comparable uncontrolled transactions undertaken in that period of two years is included in the dataset instead. The third provides that where the use of data relating to the current year "in terms of the proviso to sub-rule (5) of rule 10B" establishes that the enterprise has not undertaken a same or similar uncontrolled transaction during the current year, or that the uncontrolled transaction it undertook in the current year is not a comparable uncontrolled transaction, then irrespective of the fact that it had undertaken a comparable uncontrolled transaction in either of the two preceding financial years the price or weighted average of the prices of that enterprise's transactions shall not be included in the dataset. Sub-rule (3) prescribes the weights for that weighted average: the quantum of sales where the price was determined under rule 10B(1)(b), the quantum of costs where it was determined under rule 10B(1)(c), and "the quantum of costs incurred or sales effected or assets employed or to be employed, or as the case may be, any other base which has been considered for arriving at the respective prices" where it was determined under rule 10B(1)(e). Sub-rule (4): "Where the most appropriate method applied is a method other than the method referred to in clause (d) or clause (f) of sub-section (1) of section 92C and the dataset constructed in accordance with sub-rule (2) consists of six or more entries, an arm's length range beginning from the thirty-fifth percentile of the dataset and ending on the sixty-fifth percentile of the dataset shall be constructed and the arm's length price shall be computed in accordance with sub-rule (5) and sub-rule (6)." Sub-rule (5): "If the price at which the international transaction or the specified domestic transaction has actually been undertaken is within the range referred to in sub-rule (4), then, the price at which such international transaction or the specified domestic transaction has actually been undertaken shall be deemed to be the arm's length price." Sub-rule (6): "If the price at which the international transaction or the specified domestic transaction has actually been undertaken is outside the arm's length range referred to in sub-rule (4), the arm's length price shall be taken to be the median of the dataset." Sub-rule (7): "In a case where the provisions of sub-rule (4) are not applicable, the arm's length price shall be the arithmetical mean of all the values included in the dataset: Provided that, if the variation between the arm's length price so determined and price at which the international transaction or specified domestic transaction has actually been undertaken does not exceed such percentage not exceeding three per cent of the latter, as may be notified by the Central Government in the Official Gazette in this behalf, the price at which the international transaction or specified domestic transaction has actually been undertaken shall be deemed to be the arm's length price." Sub-rule (8): "For the purposes of this rule,— (a) 'the thirty-fifth percentile' of a dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least thirty-five per cent of the values included in the dataset are equal to or less than such value : Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the thirty-fifth percentile shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset; (b) 'the sixty-fifth percentile' of a dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least sixty five per cent of the values included in the dataset are equal to or less than such value: Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the sixty-fifth percentile shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset; (c) 'the median' of the dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least fifty per cent of the values included in the dataset are equal to or less than such value : Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the median shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset." The footnote on the /w/rule-10ca page reads: "83. Inserted by the IT (Sixteenth Amdt.) Rules, 2015, w.e.f. 19-10-2015."
Not a judgment. The statutory position is that where the application of the most appropriate method results in more than one price, rule 10CA(2) requires a dataset in ascending order; that under rule 10CA(4) an arm's length range from the thirty-fifth percentile to the sixty-fifth percentile is constructed only where the method is other than the profit split method in s.92C(1)(d) or the other method in s.92C(1)(f) and the dataset consists of six or more entries; that under rule 10CA(5) a price within that range is deemed to be the arm's length price; that under rule 10CA(6) a price outside that range takes the median of the dataset as the arm's length price; that under rule 10CA(7), in a case where sub-rule (4) is not applicable — which includes a dataset of fewer than six entries — the arm's length price is the arithmetical mean of all the values in the dataset, with a proviso deeming the price actually charged to be the arm's length price where the variation does not exceed such percentage not exceeding three per cent as may be notified by the Central Government in the Official Gazette; and that rule 10CA(8) defines the thirty-fifth percentile, the sixty-fifth percentile and the median each as the lowest value in the ascending dataset such that at least the stated proportion of the values are equal to or less than that value, with a proviso averaging that value with the value immediately succeeding it where the number of values equal to or less than it is a whole number.
Not a judgment; no judicial reasoning is stated for the section.
(a) "the thirty-fifth percentile" of a dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least thirty-five per cent of the values included in the dataset are equal to or less than such value : Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the thirty-fifth percentile shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset;
Upload it and we will read it, work out your deadline and draft the reply. A CA reviews before anything is filed.
Handle my notice → Ask a CA on WhatsAppThe range is built only where the dataset has six or more entries and the most appropriate method is not the profit split method or the other method. Rule 10CA(2) requires a dataset constructed by placing the prices in ascending order. Rule 10CA(4) then provides that where the most appropriate method applied is a method other than the method referred to in clause (d) or clause (f) of s.92C(1) and the dataset consists of six or more entries, "an arm's length range beginning from the thirty-fifth percentile of the dataset and ending on the sixty-fifth percentile of the dataset shall be constructed". Rule 10CA(5): if the price actually charged is within that range it "shall be deemed to be the arm's length price" — no adjustment. Rule 10CA(6): if it is outside the range, "the arm's length price shall be taken to be the median of the dataset". The median is therefore reached only on the footing that the range was properly constructed and the price fell outside it; there is no route to the median in a case where sub-rule (4) never applied. Where sub-rule (4) does not apply — fewer than six entries, or the profit split method or the other method — sub-rule (7) revives the older machinery: the arm's length price is the arithmetical mean of all the values in the dataset, subject to the proviso that if the variation does not exceed such percentage not exceeding three per cent as may be notified, the price actually charged is deemed to be the arm's length price. The thirty-fifth percentile, the sixty-fifth percentile and the median are all defined in sub-rule (8) on the same pattern, and the pattern is not the ordinary statistical one: each is the LOWEST value in the ascending dataset such that at least thirty-five, sixty-five or fifty per cent of the values are equal to or less than that value, with a proviso that if the number of values equal to or less than that value is a whole number, the percentile is the arithmetic mean of that value and the value immediately succeeding it. This was decided by the CBDT Circulars & Instructions (Not applicable — statutory text) and bears on section Rule 10CA, section Rule 10CA(2), section Rule 10CA(3), section Rule 10B, section Rule 10B(2), section Rule 10B(3), section Rule 10B(4), section Rule 10C(2), section 92C, section 92C(1), section 92C(2), section 92CA, section 92CA(3), section 92B, section 92BA of the Income Tax Act 1961. It is reported as Rule 10CA of the Income-tax Rules, 1962, heading "Computation of arm's length price in certain cases", transcribed from https://www.incometaxindia.gov.in/w/rule-10ca (no "Year:" stamp; the page prints "Upload Date: 13/12/2025") and corroborated on https://incometaxindia.gov.in/Rules/Income-Tax%20Rules/ITRule10CA.htm (no date stamp of any kind); rule 10B read on https://www.incometaxindia.gov.in/w/rule-10b for the methods referred to in sub-rules (2) and (3) of rule 10CA, and s.92C(1) clauses (d) and (f) read on https://www.incometaxindia.gov.in/w/section-92c-24 (Year: 2025). Three things go wrong on these facts and all three are worth taking. First, the six-entry threshold is jurisdictional to the range. An officer who has five comparables left after exclusions cannot construct a range and cannot adopt the median: he is in sub-rule (7) and must take the arithmetical mean, with the notified variation available on top of it. That is a materially different number and it is the single most common error on a small comparables set. Second, sub-rule (4) is disapplied not only by the size of the dataset but by the method — the profit split method in s.92C(1)(d) and the other method in s.92C(1)(f) are excluded from the range machinery altogether. Third, the sub-rule (8) definitions do not produce the percentile an ordinary spreadsheet function produces. The rule works by counting values that are equal to or less than a candidate value and taking the lowest value at which the count reaches the threshold; the proviso then averages that value with its immediate successor, but only where the COUNT of values equal to or less than it is a whole number. A percentile computed by interpolation, which is what most software does, will not match, and a margin that the officer's spreadsheet puts outside the range may be inside it on the words of the rule. The library already holds Barracuda Networks on the use of multiple year data in the same rule; this entry supplies the text that decision is applied to. If it applies to you, the first step is this: Count the dataset entries before anything else. Fewer than six, and rule 10CA(4) does not apply at all, there is no arm's length range, there is no median, and the arm's length price is the arithmetical mean under rule 10CA(7).
Rule 10CA, as printed on the departmental page, reads in material part: "(1) Where in respect of an international transaction or a specified domestic transaction, the application of the most appropriate method referred to in sub-section (1) of section 92C results in determination of more than one price, then the arm's length price in respect of such international transaction or specified domestic transaction shall be computed in accordance with the provisions of this rule. (2) A dataset shall be constructed by placing the prices referred to in sub-rule (1) in an ascending order and the arm's length price shall be determined on the basis of the dataset so constructed" — followed by three provisos. The first governs "a case referred to in clause (i) of sub-rule (5) of rule 10B, where the comparable uncontrolled transaction has been identified on the basis of data relating to the current year" and the comparable enterprise has in either or both of the two financial years immediately preceding the current year undertaken the same or similar comparable uncontrolled transaction, in which case the weighted average of the prices computed in the manner provided in sub-rule (3) is included in the dataset instead of the price referred to in sub-rule (1). The second governs "a case referred to in clause (ii) of sub-rule (5) of rule 10B, where the comparable uncontrolled transaction has been identified on the basis of the data relating to the financial year immediately preceding the current year" and the comparable enterprise has in the financial year immediately preceding those two financial years undertaken the same or similar comparable uncontrolled transaction, in which case the weighted average of the prices of the comparable uncontrolled transactions undertaken in that period of two years is included in the dataset instead. The third provides that where the use of data relating to the current year "in terms of the proviso to sub-rule (5) of rule 10B" establishes that the enterprise has not undertaken a same or similar uncontrolled transaction during the current year, or that the uncontrolled transaction it undertook in the current year is not a comparable uncontrolled transaction, then irrespective of the fact that it had undertaken a comparable uncontrolled transaction in either of the two preceding financial years the price or weighted average of the prices of that enterprise's transactions shall not be included in the dataset. Sub-rule (3) prescribes the weights for that weighted average: the quantum of sales where the price was determined under rule 10B(1)(b), the quantum of costs where it was determined under rule 10B(1)(c), and "the quantum of costs incurred or sales effected or assets employed or to be employed, or as the case may be, any other base which has been considered for arriving at the respective prices" where it was determined under rule 10B(1)(e). Sub-rule (4): "Where the most appropriate method applied is a method other than the method referred to in clause (d) or clause (f) of sub-section (1) of section 92C and the dataset constructed in accordance with sub-rule (2) consists of six or more entries, an arm's length range beginning from the thirty-fifth percentile of the dataset and ending on the sixty-fifth percentile of the dataset shall be constructed and the arm's length price shall be computed in accordance with sub-rule (5) and sub-rule (6)." Sub-rule (5): "If the price at which the international transaction or the specified domestic transaction has actually been undertaken is within the range referred to in sub-rule (4), then, the price at which such international transaction or the specified domestic transaction has actually been undertaken shall be deemed to be the arm's length price." Sub-rule (6): "If the price at which the international transaction or the specified domestic transaction has actually been undertaken is outside the arm's length range referred to in sub-rule (4), the arm's length price shall be taken to be the median of the dataset." Sub-rule (7): "In a case where the provisions of sub-rule (4) are not applicable, the arm's length price shall be the arithmetical mean of all the values included in the dataset: Provided that, if the variation between the arm's length price so determined and price at which the international transaction or specified domestic transaction has actually been undertaken does not exceed such percentage not exceeding three per cent of the latter, as may be notified by the Central Government in the Official Gazette in this behalf, the price at which the international transaction or specified domestic transaction has actually been undertaken shall be deemed to be the arm's length price." Sub-rule (8): "For the purposes of this rule,— (a) 'the thirty-fifth percentile' of a dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least thirty-five per cent of the values included in the dataset are equal to or less than such value : Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the thirty-fifth percentile shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset; (b) 'the sixty-fifth percentile' of a dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least sixty five per cent of the values included in the dataset are equal to or less than such value: Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the sixty-fifth percentile shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset; (c) 'the median' of the dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least fifty per cent of the values included in the dataset are equal to or less than such value : Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the median shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset." The footnote on the /w/rule-10ca page reads: "83. Inserted by the IT (Sixteenth Amdt.) Rules, 2015, w.e.f. 19-10-2015." The matter was decided on 2015-10-19 by the CBDT Circulars & Instructions (Not applicable — statutory text). On those facts the CBDT Circulars & Instructions held as follows. Not a judgment. The statutory position is that where the application of the most appropriate method results in more than one price, rule 10CA(2) requires a dataset in ascending order; that under rule 10CA(4) an arm's length range from the thirty-fifth percentile to the sixty-fifth percentile is constructed only where the method is other than the profit split method in s.92C(1)(d) or the other method in s.92C(1)(f) and the dataset consists of six or more entries; that under rule 10CA(5) a price within that range is deemed to be the arm's length price; that under rule 10CA(6) a price outside that range takes the median of the dataset as the arm's length price; that under rule 10CA(7), in a case where sub-rule (4) is not applicable — which includes a dataset of fewer than six entries — the arm's length price is the arithmetical mean of all the values in the dataset, with a proviso deeming the price actually charged to be the arm's length price where the variation does not exceed such percentage not exceeding three per cent as may be notified by the Central Government in the Official Gazette; and that rule 10CA(8) defines the thirty-fifth percentile, the sixty-fifth percentile and the median each as the lowest value in the ascending dataset such that at least the stated proportion of the values are equal to or less than that value, with a proviso averaging that value with the value immediately succeeding it where the number of values equal to or less than it is a whole number.
Not a judgment; no judicial reasoning is stated for the section. In the words reproduced by the source cited on this page: "(a) "the thirty-fifth percentile" of a dataset, having values arranged in an ascending order, shall be the lowest value in the dataset such that at least thirty-five per cent of the values included in the dataset are equal to or less than such value : Provided that, if the number of values that are equal to or less than the aforesaid value is a whole number, then the thirty-fifth percentile shall be the arithmetic mean of such value and the value immediately succeeding it in the dataset;"
It was decided by the CBDT Circulars & Instructions on 2015-10-19 and is reported as Rule 10CA of the Income-tax Rules, 1962, heading "Computation of arm's length price in certain cases", transcribed from https://www.incometaxindia.gov.in/w/rule-10ca (no "Year:" stamp; the page prints "Upload Date: 13/12/2025") and corroborated on https://incometaxindia.gov.in/Rules/Income-Tax%20Rules/ITRule10CA.htm (no date stamp of any kind); rule 10B read on https://www.incometaxindia.gov.in/w/rule-10b for the methods referred to in sub-rules (2) and (3) of rule 10CA, and s.92C(1) clauses (d) and (f) read on https://www.incometaxindia.gov.in/w/section-92c-24 (Year: 2025). Binding on the department, not on the assessee or the courts. An assessee may rely on a circular that is beneficial to them. A CBDT circular or instruction binds officers of the department but not the assessee and not the courts. Where a circular helps you, you may hold the department to it. Where it hurts you, it cannot override the Act or a judgment. On section Rule 10CA, section Rule 10CA(2), section Rule 10CA(3), section Rule 10B, section Rule 10B(2), section Rule 10B(3), section Rule 10B(4), section Rule 10C(2), section 92C, section 92C(1), section 92C(2), section 92CA, section 92CA(3), section 92B, section 92BA, the practical question is whether the facts of your own notice match the facts of this case closely enough for the same rule to apply.
It cuts both ways and is cited by both sides. Not a judgment. The statutory position is that where the application of the most appropriate method results in more than one price, rule 10CA(2) requires a dataset in ascending order; that under rule 10CA(4) an arm's length range from the thirty-fifth percentile to the sixty-fifth percentile is constructed only where the method is other than the profit split method in s.92C(1)(d) or the other method in s.92C(1)(f) and the dataset consists of six or more entries; that under rule 10CA(5) a price within that range is deemed to be the arm's length price; that under rule 10CA(6) a price outside that range takes the median of the dataset as the arm's length price; that under rule 10CA(7), in a case where sub-rule (4) is not applicable — which includes a dataset of fewer than six entries — the arm's length price is the arithmetical mean of all the values in the dataset, with a proviso deeming the price actually charged to be the arm's length price where the variation does not exceed such percentage not exceeding three per cent as may be notified by the Central Government in the Official Gazette; and that rule 10CA(8) defines the thirty-fifth percentile, the sixty-fifth percentile and the median each as the lowest value in the ascending dataset such that at least the stated proportion of the values are equal to or less than that value, with a proviso averaging that value with the value immediately succeeding it where the number of values equal to or less than it is a whole number. It arises in Assessment & Scrutiny, How Tax Law Is Read and Evidence & Burden of Proof matters, on section Rule 10CA, section Rule 10CA(2), section Rule 10CA(3), section Rule 10B, section Rule 10B(2), section Rule 10B(3), section Rule 10B(4), section Rule 10C(2), section 92C, section 92C(1), section 92C(2), section 92CA, section 92CA(3), section 92B, section 92BA of the Income Tax Act 1961, and was decided by Not applicable — statutory text. Before relying on it, read the source linked on this page and check whether it has since been distinguished, overruled or overtaken by an amendment to the Income Tax Act. In practice the steps that follow from it are these. Check the method. If the most appropriate method is the profit split method under s.92C(1)(d) or the other method under s.92C(1)(f), rule 10CA(4) is excluded on its own words and sub-rule (7) governs however large the dataset. Recompute the thirty-fifth and sixty-fifth percentiles on the words of rule 10CA(8) and not with a spreadsheet percentile function. Arrange ascending, count the values equal to or less than each candidate, take the lowest value at which the count reaches the threshold, and apply the proviso only where that count is a whole number. If the price is inside the range, rule 10CA(5) deems it to be the arm's length price and the enquiry ends; put that in terms rather than arguing quantum. Where the officer has gone to the median, make him show the constructed range first. The median under rule 10CA(6) is available only because the price was outside the range referred to in sub-rule (4), and if sub-rule (4) was never engaged the median has no statutory basis.
Still good law. The text is current so far as I could establish, but the qualification matters: NEITHER departmental page for rule 10CA carries a "Year:" stamp, so the dating test the brief prescribes for Act pages cannot be run on this rule. The /w/rule-10ca page prints "Upload Date: 13/12/2025" and the Rules-directory page ITRule10CA.htm prints no date at all. The two pages agree on every operative sub-rule and on the numbering, which is the strongest corroboration available from the departmental site; they differ only in that the Rules-directory page notes, against the defined term in sub-rule (8)(b), that the word "sixth" is used in the Gazette copy. The only amending instrument printed is footnote 83, "Inserted by the IT (Sixteenth Amdt.) Rules, 2015, w.e.f. 19-10-2015", and I could not establish from either page whether the rule has been amended since 2015. The notified percentage referred to in the proviso to sub-rule (7) is not stated in this entry; it is in the companion entry on the provisos to s.92C(2). I carried out no check of judicial treatment of rule 10CA on this pass. No source could be cited for that finding. Checking whether an authority still stands matters as much as knowing what it held: a decision may be overruled on one point and survive on another, or the provision it interprets may have been amended since. Read the source and the editor's note on this page before relying on it in a reply to an Assessing Officer or in an appeal.
Rule 10CA was read on two departmental pages and they agree on the numbering and on every operative sub-rule. https://www.incometaxindia.gov.in/w/rule-10ca prints "Upload Date: 13/12/2025" and carries NO "Year:" stamp; https://incometaxindia.gov.in/Rules/Income-Tax%20Rules/ITRule10CA.htm carries no date stamp of any kind. NEITHER RULE PAGE ON THE DEPARTMENTAL SITE CARRIES A YEAR STAMP — the Year stamp convention is used on the /w/section-<n>-<k> Act pages and not on the Rules pages, so the dating discipline that works for sections does not work for rules and I record that rather than pretend otherwise. THE SUB-RULE NUMBERING, stated because the brief for this entry put the percentile definitions in sub-rules (4) and (5): both departmental pages print the range in sub-rule (4), the within-the-range deeming in sub-rule (5), the median in sub-rule (6), the arithmetical mean in sub-rule (7), and the DEFINITIONS of the thirty-fifth percentile, the sixty-fifth percentile and the median in sub-rule (8), clauses (a), (b) and (c). The percentile definitions are in sub-rule (8), not (4) and (5). THE GAZETTE COPY AND THE DEPARTMENTAL COPY DIFFER ON A WORD IN SUB-RULE (8). The Rules-directory page ITRule10CA.htm prints the defined term in clause (b) of sub-rule (8) with an asterisk — "the *sixty-fifth percentile" — and the asterisk carries the note, printed on that page: "*Word 'sixth' is used in Gazette copy." So the Gazette text of clause (b) reads "sixth" where the departmental text reads "fifth", in the defined term that sub-rule (4) then uses to fix the upper end of the arm's length range. The two copies of the rule are not word-identical at that point. Nothing turns on it for the computation — sub-rule (4), the proviso to clause (b) and the /w/rule-10ca page all use "the sixty-fifth percentile", and clause (b)'s own operative words on both pages are "at least sixty five per cent of the values", unhyphenated as printed — but a reader comparing the Gazette against the departmental copy needs to know the discrepancy is there and is noted by the Department itself. HEALTH WARNING ON THAT CLAUSE: across three separate fetches of ITRule10CA.htm the asterisked defined term came back rendered three different ways, and on an earlier pass this library recorded it as a contradiction between the two departmental pages. It is not a contradiction; the two pages agree, and the asterisk is a footnote marker pointing to the Gazette note quoted above. Clause (b)'s opening words should be read off the page itself rather than relied on from any transcription, including this one. FOOTNOTES: /w/rule-10ca prints one footnote, "83. Inserted by the IT (Sixteenth Amdt.) Rules, 2015, w.e.f. 19-10-2015.", and ITRule10CA.htm carries the asterisked Gazette note set out above. `decided_on` is 2015-10-19 on that footnote. I did not read the IT (Sixteenth Amdt.) Rules, 2015 themselves and give the amending instrument as the footnote prints it. I did NOT transcribe the three numerical illustrations printed at the foot of ITRule10CA.htm and state nothing about them. The tolerance percentage referred to in the proviso to sub-rule (7) is not in the rule; it is notified, and it is dealt with in the companion entry on the second and third provisos to s.92C(2). This library shows the verification state of every entry openly. This entry has not yet been read in full by a chartered accountant. The summary reflects the sources listed on this page. Read the source before you rely on it in a reply to an Assessing Officer or in an appeal before the Commissioner (Appeals) or the Income Tax Appellate Tribunal.
Not a judgment. The statutory position is that where the application of the most appropriate method results in more than one price, rule 10CA(2) requires a dataset in ascending order; that under rule 10CA(4) an arm's length range from the thirty-fifth percentile to the sixty-fifth percentile is constructed only where the method is other than the profit split method in s.92C(1)(d) or the other method in s.92C(1)(f) and the dataset consists of six or more entries; that under rule 10CA(5) a price within that range is deemed to be the arm's length price; that under rule 10CA(6) a price outside that range takes the median of the dataset as the arm's length price; that under rule 10CA(7), in a case where sub-rule (4) is not applicable — which includes a dataset of fewer than six entries — the arm's length price is the arithmetical mean of all the values in the dataset, with a proviso deeming the price actually charged to be the arm's length price where the variation does not exceed such percentage not exceeding three per cent as may be notified by the Central Government in the Official Gazette; and that rule 10CA(8) defines the thirty-fifth percentile, the sixty-fifth percentile and the median each as the lowest value in the ascending dataset such that at least the stated proportion of the values are equal to or less than that value, with a proviso averaging that value with the value immediately succeeding it where the number of values equal to or less than it is a whole number.
TaxSphere, “Statutory position — rule 10CA: the dataset, the arm's length range from the thirty-fifth to the sixty-fifth percentile where there are six or more entries, the median where the price falls outside it, and the arithmetical mean under sub-rule (7) where there are fewer than six”, https://taxnotice.vittsphere.com/caselaw/case/statutory-position-rule-10ca-arms-length-range-thirty-fifth-to-sixty-fifth-percentile-and-the-median/ (validity last checked 2026-09-17)
The judgment itself is a government work and may be quoted freely. The summary, the validity note and the reasoning on this page are this library's own writing: quote them with attribution, and please do not present either as the words of the court — this page keeps the two apart and so should a quotation of it.
Every entry in this library links to where it was found, so you can check it yourself rather than take our word for it.
The Transfer Pricing Officer has refused our working capital adjustment on the ground that no rule provides for it. Where in the Rules does the right to a working capital or risk adjustment come from, and does anything prescribe how it is computed?
The Transfer Pricing Officer's adjustment is within three per cent of our price. Can I still claim the tolerance band under the proviso to s.92C(2), or has that gone?
The TPO has taken a three-year weighted average margin for a comparable under Rule 10CA, but the company fails my turnover filter in the two earlier years. Must those years still go into the weighted average?
The Transfer Pricing Officer has thrown out my working capital adjustment saying there is no prescribed method for it. Is he right that there is no prescribed method, and if he is, what does that do to the claim?
