Reading the Per-Item and Per-Category Duty Rules Correctly
Data this note rests on: Three items of the same category count as one assessment unit under the per-category reading and as three under the per-item reading, while the value of the basket stays identical at $66.69 for three median T-shirts or $206.04 for three median pairs of shoes.
What the text says
The same basket contains one unit or three units depending on which reading of the duty text is applied. In public discussion of the European rules, two assessment conventions appear side by side: one counts each item separately, and one counts a category as a single assessment unit. Under the second reading, three shirts of the same kind form one unit rather than three. Under the first, they form three. Nothing about the shirts changes between the readings; only the count of units does.
The distinction between variants matters here, so it is worth stating precisely. Different colours or sizes of the same product are normally treated as the same category, which means that splitting a basket into one shirt per colour does not create three categories. The category is a property of what the goods are, not of how they were listed or how they were packed, and repacking a parcel does not change it.
What the counting unit changes in money depends entirely on the form of the charge above the threshold, and this is where most readings go wrong. Where the charge is a percentage applied to the total value of the consignment, the two readings converge on the same number, because one unit worth the sum of three items and three units worth the same sum produce the same percentage. Where the charge is a fixed amount per assessed unit, the readings diverge by a factor of three. Where a threshold is tested per unit rather than per consignment, the readings can diverge on whether the charge applies at all.
| Question | Per-item reading | Per-category reading |
|---|---|---|
| Units in a basket of three same-category items | Three units | One unit |
| Treatment of a different colour of the same product | Usually the same category, so still one category | Usually the same category, so still one category |
| Effect where the charge is a percentage of total value | Same total | Same total, because the value is unchanged |
| Effect where the charge is a fixed amount per unit | Three times the per-unit amount | One times the per-unit amount |
| Effect on a threshold tested per consignment | No difference | No difference |
| Effect on a threshold tested per unit | Each unit tested against the line | The category tested against the line as one figure |
- Source:
- Public discussion of European per-category assessment, read against the charge types held in the six-destination parameter table on this site.
- Sample:
- No qualifying sample yet. No ruling text, no assessment notice and no clearance record is held, so the table compares readings rather than observed assessments.
- Recorded:
- 2026-W40
- Known gap:
- Which reading applies to a given parcel is not stated here, because that depends on the text in force and on the goods. The table describes what each reading does to a count, not which one a given authority applies.
This site publishes no clearance rate, no probability of assessment and no prediction of what any authority will do with a parcel. A rate would need a sample of assessed parcels with their declarations and their outcomes. That sample does not exist here, so the honest output of this article is the arithmetic of the two readings and a statement of what remains unknown.
Who is affected
The reading changes a total only when three conditions hold at once. The basket must contain more than one item of the same category, the destination must apply a per-unit amount or test a threshold per unit rather than per consignment, and the basket must sit close enough to a line that the count of units matters. A single-item parcel is unaffected. A basket spread across categories is largely unaffected, because each category contributes its own unit either way. A basket assessed purely as a percentage of its total value is unaffected, because the value is the same number under both readings.
That leaves a specific kind of order in scope: several items of one kind, at a destination where a fixed amount or a per-unit test exists, with a total near a line. Repeated purchases of the same product are the typical shape, which is exactly the pattern that haul building encourages, because buying three of something is usually cheaper per item and easier to consolidate than buying three different things.
| Situation | Does the reading change the total | Why |
|---|---|---|
| One item in the parcel | No | A basket of one is one unit under either reading. |
| Three items from three different categories | Usually no | Each category forms its own unit under either reading, so the count matches. |
| Three items of one category, percentage charge only | No | The base is the value of the goods, which does not change with the count. |
| Three items of one category, fixed amount per unit | Yes | One unit against three units multiplies the per-unit amount. |
| Three items of one category, threshold tested per unit | Yes | Three tests against a line can produce a different outcome from one test. |
| Three items of one category, threshold tested per consignment | No | The comparison uses the consignment figure, which is the same either way. |
| A basket spread across categories near a line | Possibly | The unit count changes only for the categories that repeat within the basket. |
- Source:
- The charge types and threshold logic published with the landed cost instrument, combined with the two readings described above.
- Sample:
- No qualifying sample yet. The table classifies situations rather than observations, and no parcel in the entry pool was assessed by any authority.
- Recorded:
- 2026-W40
- Known gap:
- Whether a destination tests its threshold per unit or per consignment is not confirmed by any document held here. That single unknown decides several rows above, and it is left open rather than assumed.
One practical consequence is worth naming before the worked examples. Splitting a parcel does not change what the goods are, so it does not change the category count under a per-category reading; it changes the number of consignments, which interacts with a per-consignment threshold test and with freight and base costs. Grouping and splitting are therefore decisions about consignments, while the per-item and per-category question is a decision about units inside one consignment. Confusing the two is the most common error in discussions of this rule.
Worked effect per category
The table below builds a three-item basket in each of the eight categories, using the recorded category median from the pool extracted on 2026-09-29. The basket totals are arithmetic on those medians, not observed orders. The value of each basket is identical under both readings, which is the point of the table: what changes is the number of units, and with it the effect of any per-unit amount or per-unit threshold test.
| Category | Median | Three-item basket | Units per-category | Units per-item | Where the difference shows |
|---|---|---|---|---|---|
| Shoes | $68.68 | $206.04 | 1 | 3 | A high-value basket, so a percentage charge is large and a per-unit amount is smaller by comparison. |
| Jackets | $53.89 | $161.67 | 1 | 3 | Sits above several destination lines on its own, so the reading changes the unit count rather than whether a line is crossed. |
| Pants and Shorts | $44.62 | $133.86 | 1 | 3 | Close to the European line in value terms, which makes the unit count the deciding factor. |
| Hoodies and Sweaters | $37.18 | $111.54 | 1 | 3 | The category with the most entries in the pool, so three of one kind is a realistic basket. |
| Other Items | $35.81 | $107.43 | 1 | 3 | A mixed category, where three items may not share a category in practice even when the median is used here. |
| Accessories | $26.77 | $80.31 | 1 | 3 | The widest category in the pool, so a median-built basket represents the middle and not a typical order. |
| T-Shirts | $22.23 | $66.69 | 1 | 3 | A low-value basket, where a fixed per-unit amount is a larger share of the total than anywhere else in the table. |
| Headwear | $16.22 | $48.66 | 1 | 3 | The cheapest basket here, so any fixed amount per unit is proportionally the heaviest. |
- Source:
- Category medians from the entry pool extracted on 2026-09-29, multiplied by three to build a comparable basket per category.
- Sample:
- n = 8 category medians drawn from a 195-entry pool, with 24 to 26 entries behind each median. The three-item baskets are constructed, not observed.
- Recorded:
- 2026-W40
- Known gap:
- Basket totals are arithmetic on recorded prices in US dollars, while destination thresholds are stated in local currency, so a comparison depends on the rate applied on the day. No conversion is applied here, and no basket in the pool was assessed by any authority.
Two rows carry more of the argument than the rest. The T-Shirts row produces the smallest basket at $66.69, which means a fixed amount charged per unit is proportionally largest there: under a per-item reading a modest per-unit amount can add a noticeable share to a cheap basket, and under a per-category reading it is charged once. The Shoes row produces the largest basket at $206.04, where a percentage charge dominates and the unit count matters less to the total, though it may still matter to whether a line is crossed at all.
The Accessories row is the caution in the table. Its median of $26.77 comes from a category whose recorded range runs from $5.17 to $451.10, so a basket built from the median could be a fraction of a real three-item order in that category. A worked example built on a median is a way to compare categories with each other, and it is not a forecast of any particular order. The same caveat applies to every row, with the Accessories row making it most visible.
What to do now
The sequence below turns the two readings into decisions that can be made before an order is placed. Each step is about establishing what kind of rule is in front of you rather than about predicting an outcome, because the second is not something this site can do.
- Establish the form of the charge at the destination: a percentage of value, a fixed amount per unit, or a combination. The six-destination parameter table carries the rate and the threshold for each row, and the landed cost instrument at /instrument/landed-cost/ applies them to a real basket.
- Establish whether the threshold is tested per consignment or per unit. This single question decides whether the category count matters to the tax line or only to a fee schedule.
- Count the repeated categories in the basket rather than the items. Three items from three categories behave like three categories under both readings; three items from one category do not.
- Group or split for the right reason. Splitting changes consignments, freight and fixed fees, and it does not change what the goods are. The reading of the European rule in /field-notes/eu-per-category-duty-rule/ covers the regrouping question in more detail.
- Keep the invoice and the payment record with the parcel. Whatever reading applies, the declared figure should match the amount paid, and the evidence for that is a document rather than a memory. The declare value question is taken apart separately in /field-notes/declare-value-question/.
- Do not plan around a colour or variant split. Different colours of the same product normally count as one category, so separating them does not separate the units.
The reading of a rule is not a prediction of an assessment. A basket that looks favourable under one reading can still be assessed under the other, and nothing on this site estimates how often either happens. A reader who needs certainty about a specific parcel needs the text in force and, where the value justifies it, professional advice rather than a calculator.
What can be done with confidence is narrower and still useful: know which reading changes the total, know how many units the basket contains under each, and know the two unknowns that remain. The first is whether the destination charges per unit. The second is how the threshold is tested. Both are answerable from the text in force for a given destination, and neither is answerable from the data held here, which is why they are written down as questions rather than filled in with assumptions.