EU trade defence stayed high in 2025, and three cases carried retroactive duties

AI Summary

The European Commission’s 2025 trade defence report records 32 new investigations, 26 definitive measures imposed and 232 measures in force at year end, 220 of them definitive. Provisional anti-dumping duties went on in 27 cases. In three cases the Commission applied measures retroactively, backdated to before the provisional stage. The report is aggregate: it names no product and no rate. The cost to an importer is therefore a risk to audit, not a rate to price, and the action is a product-line review plus a duty-risk clause in new contracts.

What happened

The Commission adopted its 2025 Annual Trade Defence Report, and the news release accompanying it was published on 18 September 2026. The release gives the year in six numbers.

It opened 32 new trade-defence investigations in 2025. That is just slightly below the record high set in 2024, so the pipeline of new cases did not fall away after the peak year.

It imposed 26 definitive trade defence measures during 2025. A definitive measure is the end state of a case: the duty that stays in force for its full term rather than the provisional duty that holds the position while an investigation finishes.

By the end of 2025 the Commission had 232 trade defence measures in place, and 220 of those were definitive. The release states that these measures protect around 637,000 manufacturing jobs in the European Union.

Provisional duties were imposed in 27 anti-dumping cases in 2025. That is the stage that matters most to an importer with cargo on the water, because a provisional duty starts the clock on money owed at the border.

And in three cases the Commission applied measures retroactively, with the measures backdated to before the introduction of provisional measures.

Retroactive duty — a duty whose legal effect starts before the date it was imposed. In EU trade defence the ordinary sequence is an investigation, then a provisional duty, then a definitive duty. A retroactive measure reaches back past the provisional stage, so goods that entered before the provisional duty existed become liable for the duty as well. The importer’s declaration date stops being the line that separates duty-free cargo from dutiable cargo.

The release is a summary, not a schedule. It does not name the three cases, the products, the countries of origin, the rates or the backdated periods. It does not name the 27 anti-dumping cases either. What it gives is the shape of the year: a heavy investigation load, a heavy definitive output, and a small but non-zero number of cases where the duty reached backwards.

Why now

The report covers calendar 2025 and was released on 18 September 2026, so the year it describes is closed and the numbers are final rather than running totals. The 32 investigations are the ones that will produce provisional and definitive measures through 2026 and 2027. The 26 definitive measures are already at the border. The three retroactive cases are already settled, and any exposure they created has already been incurred by whoever held the goods.

So what

The judgement is short. This report does not let an importer calculate a per-unit duty, because it carries no product, no origin, no rate and no period. What it does is tell an importer that the population of duty-bearing lines is large and stable, that new lines keep being added at close to a record pace, and that a small number of cases each year reach backwards past the provisional stage.

That last point is the one that changes behaviour. A provisional duty is visible. It is published, it has a start date, and an importer who reads the Official Journal knows when the border cost begins. A retroactive measure removes that certainty after the fact. Cargo that cleared at a zero rate becomes cargo that owed duty, and the bill arrives after the goods have been sold.

Work the arithmetic on the exposure rather than the rate. Take an importer with an annual import programme of EUR 4,000,000 in CIF value across a set of product lines. The report gives no rate, so the exercise is to ask what a duty would cost at each level rather than to assert one.

Those are arithmetic, not forecasts. The point of the table is the ratio: a duty in the range the EU routinely sets turns a thin import margin into a loss on the affected lines, and the affected lines are the ones an importer has not checked.

Now narrow it to the retroactive question. Suppose 10% of that programme, EUR 400,000 of CIF value, sits in a product line that later becomes one of the three retroactive cases. At a 20% duty the backdated bill is EUR 80,000. At 30% it is EUR 120,000. That is the amount that arrives with no warning, on goods already cleared and already sold, and it is the amount a duty-risk clause is written to move to the other side of the contract.

The second worked case is the in-transit one. An importer with a EUR 250,000 consignment on the water when a provisional duty is published faces the duty from the date the measure takes effect, and the report gives no date. If that consignment is caught at 20%, the cost is EUR 50,000 against a margin that was calculated without it. The operational answer is not to predict the date but to know which of your lines are in an open case, so the consignment is either accelerated, diverted or priced with the duty inside it.

The third case is the standing stock. Of the 232 measures in place, 220 are definitive. A definitive measure runs for its full term, so a product line that has been dutiable for years stays dutiable. An importer who has bought the same line from the same origin for three years without checking the measure list is carrying a cost that is already in the price and is not visible on the invoice as a separate item. Auditing the line turns a hidden cost into a negotiable one.

For the wider picture of how EU trade defence interacts with other border charges on the same tonne, the CBAM comparison in Two EU Charges Hit the Same Tonne, and Only One of Them Is Knowable Up Front sets out the difference between a charge you can price and one you cannot. The registration mechanics that precede a definitive duty are worked through in AD760: Registration Is Ordered Now, and the Money Starts in April 2027, which shows how a registration step creates a liability before the duty exists. And the safeguard side of the same regime, where tariff-rate quotas rather than anti-dumping duties set the border cost, is covered in EU safeguard tariff-rate quotas hit grain-oriented electrical steel from 18 September 2026.

How confident

Confirmed. The six figures in this article come from a single document: the European Commission’s 2025 Annual Trade Defence Report news release, published 18 September 2026 and retrieved 23 October 2026. That document states 32 new investigations, 26 definitive measures imposed, 232 measures in place at year end with 220 definitive, provisional duties in 27 anti-dumping cases, and retroactive application in three cases backdated to before the provisional stage. The characterisation of a definitive measure as one that stays in force for its full term, and of a provisional measure as the interim stage, is the ordinary meaning of those terms in EU trade defence and is consistent with the release’s own usage.

Inferred. Everything about which products, origins, rates and periods are affected is inference, and it rests on nothing in the supplied document. The release names no case, so the three retroactive cases cannot be identified from it, and the 27 provisional anti-dumping cases cannot be identified either. The cost tables in this article are arithmetic on assumed programme values, not figures from any source. The claim that a retroactive measure creates an unbudgeted bill on already-cleared goods follows from the meaning of retroactivity, not from a worked example in the release. An importer who needs the actual case list has to go to the Official Journal and the Commission’s case register, which this article does not cite and does not reproduce.

What would change my mind. A published case list naming the three retroactive cases, with products, origins, rates and backdated periods, would replace the audit instruction in this article with a specific exposure calculation for the named lines. A statement from the Commission that the three cases were confined to a single product family would narrow the audit to that family. And a later report showing the retroactive count at zero for a run of years would weaken the case for a standing duty-risk clause, though it would not remove the case for auditing lines against the 232 measures already in force.

Work it out: the landed cost calculator

Different question, same method: the step-by-step guides

For you

• Importers with lines in an open investigation: pull the Commission’s case register for every product code you buy and mark which lines sit in an investigation opened in 2025 or 2026, because the 32 new cases will produce provisional duties and a provisional duty is the stage where border cost starts.
• Importers negotiating new supply contracts: add a duty-risk clause that states who bears a duty imposed after the contract date, including any retroactive element, because the three backdated cases show that the declaration date is not a reliable boundary.
• Importers holding stock bought years ago: check the 220 definitive measures against your standing product lines, because a definitive measure runs for its full term and a duty that has been in force for years is already inside your purchase price without appearing as a separate line.
• Importers with cargo in transit: confirm the entry-into-force date of any provisional measure on your line before the vessel arrives, because the report gives no date and the consignment is priced without the duty until you check.

The data point

Three cases in 2025 had measures applied retroactively, backdated to before the introduction of provisional measures.

FAQ

Does the 2025 report tell me which products face retroactive duties?

No. The news release gives the count — three cases — and nothing else about them. It does not name the products, the countries of origin, the duty rates or the periods the measures were backdated to. The same is true of the 27 anti-dumping cases in which provisional duties were imposed: the release gives the number and no case list. An importer who needs the named cases has to work from the Official Journal and the Commission’s own case register, which sit outside this document. Treat the report as a signal about the volume of activity, not as a schedule of what is dutiable.

How does a retroactive duty change what I owe at the border?

A provisional duty has a start date, and goods declared before that date are outside it. A retroactive measure reaches back past the provisional stage, so goods that entered before the provisional duty existed become liable as well. The declaration date stops being the line that separates duty-free cargo from dutiable cargo. For an importer, the practical effect is that a consignment cleared at a zero rate can later carry a bill, and the bill arrives after the goods have been sold. That is why the exposure is a contract question as much as a customs question.

Should I stop importing lines that sit in an open case?

Not necessarily. An open investigation is not a duty, and many cases close without a definitive measure. The decision is about pricing and contract terms rather than about stopping. If a line is in an open case, price the consignment with a duty inside it, or write a clause that puts the duty on the supplier if it is imposed after the contract date. The 232 measures in force at the end of 2025, 220 of them definitive, show that a large standing population of dutiable lines is normal. The task is to know which of your lines are in that population.

How we calculated this

Report figures used, all from the 2025 Annual Trade Defence Report news release:
  new investigations initiated 2025        = 32
  definitive measures imposed 2025         = 26
  measures in place, end 2025              = 232
  of which definitive                      = 220
  provisional duties, anti-dumping cases   = 27
  retroactive cases                        = 3

Share of the standing stock that is definitive:
  220 / 232 = 0.948  ->  94.8%

Retroactive cases as a share of provisional anti-dumping cases:
  3 / 27 = 0.111  ->  11.1%

Worked exposure, assumed programme (not from any source):
  annual CIF programme                     = EUR 4,000,000
  duty at 5%   -> 4,000,000 x 0.05  = EUR   200,000
  duty at 10%  -> 4,000,000 x 0.10  = EUR   400,000
  duty at 20%  -> 4,000,000 x 0.20  = EUR   800,000
  duty at 30%  -> 4,000,000 x 0.30  = EUR 1,200,000

Retroactive slice, assumed at 10% of programme:
  4,000,000 x 0.10 = EUR 400,000 of CIF value
  at 20% duty -> 400,000 x 0.20 = EUR  80,000 backdated bill
  at 30% duty -> 400,000 x 0.30 = EUR 120,000 backdated bill

In-transit consignment, assumed:
  CIF value = EUR 250,000
  at 20% duty -> 250,000 x 0.20 = EUR 50,000

The point

The report is a count, not a rate. It tells an importer that the EU opened 32 investigations, imposed 26 definitive measures and held 232 measures in force at the end of 2025, and that three cases reached backwards past the provisional stage. None of that converts into a per-unit cost, because the document names no product and no rate. What it converts into is a decision: audit your product lines against the measures already in force and the cases already open, and put a duty-risk clause in every new contract so that a duty imposed after the signature, including a backdated one, lands on the side of the deal that agreed to carry it.