At audit, the ANAF team rejects part of the comparables in the file, rebuilds the range, recalculates the margin and moves it to the median. The result: several million lei of additional taxable base, plus interest. Between such an adjustment upheld and one struck down stands, almost always, a single thing — a comparability study built correctly and defended technically.
The transfer price is the price at which one company in a group sells goods, provides services, grants loans or licenses intangible assets to another company in the same group — "related parties". Because the parties are not independent, the price is not formed on a real market but set internally; hence the risk that profit is shifted, through the price, into a lower-tax jurisdiction.
As an antidote, the law imposes the arm's-length principle: a transaction between related parties must produce the same result as one between independent partners. The domestic basis is art. 11 of the Fiscal Code, which refers expressly to the OECD Guidelines, applicable in Romania even though the country is not a member of the organisation.
Adjustments are large for three compounding reasons. First, volume: intra-group transactions are recurrent and high-value. Second, the multiplier effect: an adjustment to the margin does not hit a single invoice, but the entire flow across the open years. Third, the recalculation mechanism: when ANAF rejects the taxpayer's position, it does not "cut" a specific expense but rebuilds the "market" price and imposes the difference across the whole transaction. That is how adjustments of millions of lei arise out of a file of a few dozen pages.
The reference rule is no longer the old OPANAF 442/2016. On 2 July 2026, two orders of the President of ANAF were published in Official Gazette no. 543: no. 828/2026, which rewrites the rules of the transfer-pricing file and the adjustment/estimation procedure, replacing Order 442/2016, and no. 827/2026, which updates the advance-pricing-agreement procedure. The new file rules apply to transactions from 2026, and the procedural provisions to audits initiated after 1 January 2027.
Large taxpayers exceeding the thresholds prepare the file annually and submit it electronically, through the Virtual Private Space, within 30 working days of the corporate-tax return deadline. Until now it was presented only at the tax authority's request.
The materiality thresholds are analysed for each transaction with each related party, not on the cumulative value. For large taxpayers, the services threshold was reduced to EUR 100,000 (from 250,000), a sign of stricter documentation on the intra-group services area.
A taxpayer who has not filed the file through the Virtual Private Space may present it, at the inspection's request, within a much shorter term — 5 working days, under the new order.
The order expressly defines an incomplete file (absence of the transaction description, the functional analysis, the justification of the method, the comparability analysis). The consequence: ANAF may proceed directly to estimating the transfer prices.
Note — a formal but weak file is treated as non-existent. Under the new regime, a file submitted formally but with a poorly documented comparability study is treated as if it did not exist. And where documentation is missing or incomplete, the tax authority no longer discusses the method with the taxpayer: it estimates the price and, on adjustment, positions the result at the median of the comparison range.
From the median upward, every percentage point translates into additional taxable base. The difference between preventing and repairing is measured, here, in the quality of the benchmarking filed on time.
Art. 11 of the Fiscal Code and the OECD Guidelines recognise five methods. The most appropriate one is chosen — not the most convenient.
Directly compares the price in the related-party transaction with the price in a comparable transaction between independents. The most direct; it works for commodities and loans, where reference prices or interest rates exist.
Starts from the supplier's costs and adds a market profit margin. Suitable for toll manufacturing and for intra-group services where the provider has routine functions.
Starts from the resale price to an independent customer and deducts a market gross margin. Typical for distributors.
Compares a net-profitability indicator of the tested party with that of comparable independent companies. The most used in practice, because it needs comparables of profitability, not of product.
Allocates the combined profit of the transaction between the related parties, by each one's contribution. Used where both parties bring unique and valuable contributions and the other methods do not work.
The "tested party" is the entity whose remuneration is analysed — usually the one with the simpler, more easily comparable functions. Correctly establishing the tested party and the method conditions the whole of the rest of the file.
Benchmarking is the search, in a database of financial statements, for comparable independent companies, in order to determine the "market" profitability range. In practice, this is where the file is won or lost.
The functions performed, the assets used and the risks assumed by each party are described. It determines the profile (routine vs complex) and, implicitly, the margin due to the tested party.
The database, the activity codes, the independence criteria and the quantitative filters. The new order imposes a geographic hierarchy: first Romania, then the EU (including the United Kingdom), the EMEA region and, only where needed, the international level.
From the automatically produced set, each potential comparable is accepted or rejected manually, with justification. Here arise most of the disputes.
From the margins of the comparables a range is built; the first quartile, the median and the third quartile delimit the "market zone". If the tested party's margin falls within the range, the price is compliant; if not, the adjustment appears.
The new order requires the study to be presented also in Excel format, with the search strategy, the selection formulas, the set of comparables and the range calculation — precisely the elements an expert can verify and, if necessary, rebuild. Where the taxpayer errs: comparables chosen superficially, undocumented filters, a functional analysis that does not support the profile. Where ANAF errs: rejecting comparables wholesale without proving why they are not comparable, using data the taxpayer cannot verify, applying the median automatically without testing whether the transaction already fell within the range.
The typical mechanism of a large adjustment has three moves: the inspection rejects part of the taxpayer's comparables, narrowing the range; it rebuilds the range on a narrower set, often with higher margins; it positions the taxpayer's result at the median of the new range and imposes the difference as additional income. Each move is technically challengeable — and when the audit ends with a huge amount to pay, this is where the battle is fought.
The most disputed point is the adjustment to the median. In the new order, ANAF confirms the use of the market's central tendency — the median of the interquartile range or, in certain situations, the arithmetic mean. In practice, the tax authority usually positions the result at the median. Automatic positioning at the median is not, however, unchallengeable: it has been argued in the specialist debate that, where the taxpayer was already within the range or close to it, adjustment to the median is excessive, and the correct point would be toward the lower quartile. This is a technical question, of method — not one that depends on invoking a particular precedent — that a party-appointed expert can turn from a "difference of millions" into an "adjustment struck down".
Redoes the database search, checks the filters and the range, tests whether the ANAF rejections hold up.
Argues why the chosen method (for instance, TNMM with the tested party in Romania) is the most appropriate and why the inspection's alternative does not hold.
Shows point by point why the comparables added by the inspection are not comparable (functions, markets, different years) and why the rejected ones were valid.
Documents the real profile of the tested party — the element that decides the margin due.
If an adjustment is nonetheless justified, shows that the correct level is not the median but another point of the range, reducing the additional base.
In the audit, the expert works alongside the taxpayer on the point of view on the draft report — the moment when a well-built technical position can stop the adjustment before it becomes a tax assessment. If the adjustment is issued nonetheless, the ground shifts to administrative litigation, where the central evidence is the judicial expert report as evidence in tax litigation.
The court appoints an expert and sets the objectives — the technical questions he must answer: whether the method was correctly chosen, whether the rejection of comparables was justified, what the correct range is. The framing of the objectives is itself a stake. Art. 330 of the Code of Civil Procedure allows each party to appoint an adviser — the party-appointed expert — who attends the expert examination and formulates objections; in connected criminal files, the equivalent role is exercised under art. 172 of the Code of Criminal Procedure. A judicial report left technically unchallenged usually becomes the basis of the judgment.
The safest defence is the one that makes any adjustment unnecessary. The advance pricing agreement (APA) is the arrangement by which the taxpayer obtains, before the transactions are carried out, ANAF's confirmation of the transfer-pricing method. Once issued and observed, the APA blocks subsequent adjustments on the covered transactions.
The procedure was updated through OPANAF 827/2026, which introduces, among other things, the possibility of applying the agreement to earlier periods ("roll-back"), under certain conditions. For groups with large and recurrent flows, the APA turns a risk of millions into an administrative certainty.
A low margin in an entity without economic substance — without people, functions and real decisions — is doubly vulnerable: to a price adjustment and to reclassification of the structure. The substance must support the functional profile invoked in the file.
And certain intra-group arrangements with a cross-border tax effect fall within the DAC6 reporting of arrangements. A solid transfer-pricing file requires coherence with both: substance supports the profile, and reportable arrangements are declared.
A Romanian distribution subsidiary receives, at audit, an adjustment of several million lei. The reason invoked — an operating margin below the median of the range rebuilt by ANAF, after the inspection had rejected half the comparables in the file. On re-analysis, the party-appointed expert shows that the rejected companies were genuinely comparable (the same distribution functions, the same market), that two of the comparables added by ANAF had manufacturing functions — hence a different profile — and that, on the correctly rebuilt range, the subsidiary's margin fell between the first quartile and the median.
The technical conclusion: no adjustment was justified. It was not rhetoric that overturned the amount, but the rigorous reconstruction of the range. This is where, in transfer pricing, the fate of the adjustment is decided.
Yes, if it has transactions with related parties above the materiality thresholds. The new order analyses the thresholds for each transaction with each related party, so a small company in a large group may have a file obligation even for a single category of transactions — services, royalties, loans. Underestimating this obligation is one of the most frequent vulnerabilities.
The absence or incompleteness of the file allows ANAF to estimate the transfer prices, usually by positioning the result at the median of the range. In practice, you lose control over the method and the comparables, and the adjustment is made on the tax authority's analysis. Later reconstruction, through an expert, is possible but far harder than a file prepared in time.
Yes. The adjustment is challenged through an administrative complaint and then in court, where the judicial expert report is the decisive evidence. A party-appointed tax expert can show that the method, the comparables or the level of the adjustment (median vs lower quartile) are wrong. Many large adjustments do not survive a rigorous technical re-analysis.
It depends on the value and recurrence of the transactions. For large and repeated intra-group flows, the advance pricing agreement removes uncertainty and the risk of adjustment on the covered transactions, and from 2026 it can cover, under given conditions, earlier periods as well. For small and occasional transactions, the procedural effort may exceed the benefit.
Informational material, updated 18 July 2026. It does not constitute legal or tax advice; individual situations must be analysed case by case.
An initial technical review tests whether the method, the comparables and the median positioning hold — in the audit, where the adjustment can still be stopped, or in court, where the judicial expert report decides.