How sets and kits are classified under GRI 3(b), and how to check past entries that carried them

Where do sets fit in the General Rules of Interpretation?

Sets are a GRI 3 problem, and GRI 3 is only reached when GRI 1 and GRI 2 have not settled the classification. The order matters because several headings name sets by their terms, and a set that fits one of them never gets to an essential-character analysis.

Two of those headings carry a rule an audit should know: the set pays the highest rate among its articles, which is the opposite of what an essential-character choice can do.

Only when no heading describes the set as a whole does GRI 3 apply. GRI 3(a) normally prefers the most specific heading, but it says that when two or more headings each refer to only part of the items in a set put up for retail sale, those headings are to be regarded as equally specific. That sends the set to GRI 3(b): goods put up in sets for retail sale are classified as if they consisted of the component that gives them their essential character, "insofar as this criterion is applicable". If it is not, GRI 3(c) puts the set under the heading that occurs last in numerical order among those that equally merit consideration.

Keep sets separate from cases. Under GRI 5(a), a case specially shaped or fitted for an article, suitable for long-term use and entered with it, is classified with the article, unless the case itself gives the whole its essential character. A camera in its fitted case is not a set; a case holding a mix of different tools may be.

What are the three parts of the set test?

The legal text of GRI 3(b) does not define a set. CBP applies the definition in Explanatory Note X to the rule, quoted in rulings such as HQ H061205 (August 10, 2009). Goods put up in sets for retail sale are goods that:

1. Consist of at least two different articles that are, prima facie, classifiable in different headings. The Note's own example is six fondue forks, which are not a set.
2. Are put up together to meet a particular need or carry out a specific activity. In HQ H061205 a stainless steel infuser and a glass tumbler qualified because they were put up together to make and drink tea.
3. Are put up in a manner suitable for sale directly to users without repacking, for example in boxes or cases or on boards. In HQ H061205 the items came in a small retail cardboard box.

All three have to hold. Items packed together for shipping but repacked or sold separately in the United States fail the third part. Items that share a box but do not serve one activity fail the second. Two of the same article fail the first.

How do you find the component that gives a set its essential character?

Explanatory Note VIII, as CBP quotes it, says the factor that decides essential character varies with the goods: the nature of the material or component, its bulk, quantity, weight or value, or the role of a constituent material in relation to the use of the goods. CBP and the courts weigh all of them and look for the one that tips the balance.

In HQ H061205, CBP worked through each factor in turn. Bulk, quantity and weight did not help, because the infuser and the tumbler were about the same size and weight. Value did: the infuser was 70 percent of the cost. Role in use did too: the infuser brewed the tea, acted as a lid and gave the product its marketing name. The set went to heading 7323 as a stainless steel household article, not heading 7013 as glassware. CBP distinguished two earlier rulings where a teapot, being larger and more expensive and marketed as a teapot, gave a tea set its character instead.

In Better Home Plastics Corp. v. United States (119 F.3d 969, Fed. Cir. 1997), the set was an inexpensive shower curtain: a semi-transparent textile outer curtain, a plastic liner and plastic hooks. CBP had found that neither component gave the set its essential character and used GRI 3(c) to put it in the textile heading at 12.8 percent. The Court of International Trade found that the liner did, for its water-retaining, protective, privacy and decorative functions and the low cost of the set, which put it in the plastics heading at 3.36 percent. The Federal Circuit affirmed and rejected the fall-back to GRI 3(c).

Two lessons carry over to an audit. Essential character is a fact question, so the file needs the facts: component costs, weights, dimensions, how the set is marketed and how it is used, written down next to the heading chosen, as a classification report does. And a set's classification can move a lot of duty on a narrow factual call.

What does the choice of heading do to a stacked rate?

Take an illustrative lunch kit, product of China, entered on November 10, 2026: an insulated lunch bag with an outer surface of plastic sheeting, plus a set of plastic food containers, packed together for retail sale. 5,000 kits at $8.00, $40,000 in all, of which the bag accounts for $5.00 and the containers for $3.00 per kit. The rates come from HTS Revision 23 (2026), in force from that date. No Section 301 exclusion in force under headings 9903.88.69 or 9903.88.70 (extended through January 9, 2027) names either article, and no IEEPA duty applies, because CBP stopped collecting IEEPA duties on goods entered from February 24, 2026 (CSMS # 67834313):

• Subheading 4202.92.10 (insulated food or beverage bags with an outer surface of plastic sheeting; the textile-surface bags are in 4202.92.04 and 4202.92.08): general rate 3.4 percent, and on the U.S. note 20(f) list for Section 301 heading 9903.88.03, an additional 25 percent.
• Subheading 3924.10.40 (plastic tableware and kitchenware, other): general rate 3.4 percent; it is not on a Section 301 China list in effect, because its listing under heading 9903.88.16 is suspended.
• Both carry the Section 301 forced-labor duty for products of China, heading 9903.05.31, an additional 12.5 percent, for goods entered from July 24, 2026 (USTR notice, 91 FR 47318), since neither is in the U.S. note 52 exemptions.

The base rates are the same 3.4 percent in all three rows. The difference is the Section 301 list layer: treated as a set under the bag's subheading, the containers pick up 25 percent they would not carry alone. That is $3,750 between the first two rows on one shipment, and $10,000 between the first and the third. The merchandise processing fee does not change, because the value does not.

This is not hypothetical. CBP has classified close versions of this kit both ways. In NY I87125 (October 9, 2002), an insulated lunch bag with a plastic drink bottle and food container, presented on entry put up for retail sale as a set, was a set, with the bag giving essential character, under 4202.92.1000. In NY N303419 (April 12, 2019), a five-piece lunch bag of an insulated bag, plastic food containers and a bottle was found not to be a set or a composite good; the bag went under 4202.92.1000 with the Section 301 heading and the containers under 3924.10.4000. The 2019 ruling gives its conclusion without explaining which part of the test failed, and a ruling binds only for articles identical to the one it describes (19 CFR 177.9(b)(2)). For your own kit, the facts of the three-part test decide, and a ruling request is the way to settle them before the next shipment; see binding rulings and CROSS.

To price a code on its entry date, use the duty calculator. How the Chapter 99 layers stack on a line is in why your entry shows the wrong duty stack.

How is a set reported on Form 7501?

The Form 7501 instructions (revision 02/26) give sets their own reporting. For a set classified under GRI 3(b) or 3(c), column 33 carries the HTS number from which the set's rate is derived, preceded by the special program indicator "X", and that line carries the total value, quantity and charges of the set and all duties, taxes and fees. Each article in the set, including the one on the "X" line, is then reported on a separate line as if separately classified, preceded by "V", with the quantity and value attributed to it. Quota, visa, licensing and other agency requirements go on the "V" lines.

That reporting is what makes sets auditable from the data. A correctly reported set shows up as one "X" line with duty and a group of "V" lines with values and no separate duty. A set reported as loose lines shows up as several duty-bearing lines with no indicators.

How do you spot set errors in past entries?

Errors run both ways, so check both.

Sets entered as separate lines. These can be overpayments or underpayments, depending on which component gives the set its essential character.

1. Search the invoice descriptions in the entry export for "set", "kit", "gift box", "combo", "bundle" and "pack", and for SKUs whose retail packaging holds articles from different headings.
2. For each hit with no "X" and "V" indicators, test the three parts: different articles, one activity, packed for sale to users without repacking.
3. If it is a set, identify the essential-character component and recompute the whole value at that component's full stack. If the result is higher, the loose lines underpaid, as the lunch kit above would have; if lower, they overpaid.

Loose goods entered as a set. These are usually overpayments.

1. List every "X" line and the "V" lines under it.
2. Check whether the goods are really sold together. Items shipped in one carton and split for sale in the United States fail the third part.
3. Check whether the "V" lines include identical articles, which fail the first part, or articles with no shared purpose, which fail the second.
4. Check that the "X" component really gives the set its character. A set placed under its highest-duty component on thin facts is a common overpayment.

Sets that belong under a heading that names them. Tool sets, assorted kitchen sets and travel sets have their own headings. A set of tools of headings 8202 to 8205 classified by essential character instead of under 8206 is wrong whichever way the rate goes.

Record each conclusion with the HTS revision, the entry date and the facts used, because essential character turns on facts that are easy to lose: component costs, weights and the retail packaging.

What do you do with each finding?

The routes are the same as for any classification error. While an entry is unliquidated, a post-summary correction can be filed within 300 days of the entry date and at least 15 days before the scheduled liquidation date, whichever is earlier (CBP post-summary correction page). After liquidation, an overpayment is recovered by protest within 180 days. An underpayment that a post-summary correction cannot reach is a candidate for a prior disclosure. The full routes, the evidence for each and what a wrong code costs with interest and penalty exposure are in how to audit HTS codes on past entries and what a misclassification costs. Confirm any filing with a licensed customs broker.

If the set was designed so that a lower-duty component gives it its character, the question becomes whether the design is real; that is covered in tariff engineering or misclassification.

How does TariffSense fit into a set review?

TariffSense reads Form 7501 lines and broker entry exports, including the special program indicators, checks each line's classification and Chapter 99 stack against the measures in force on its entry date, and looks across your history for the same product classified differently from one entry to the next. On sets it reads the "X" and "V" lines and flags both directions: articles shipped and invoiced together as a retail set but entered as separate lines under their own headings, and separate articles entered as a set on an "X" line that do not read as a set, or that belong under a heading that names the set. The duty calculator prices each candidate heading for the entry date, so the effect of the set decision can be stated per line. The classification call and any correction stay with your team and your broker. The post-entry audit page describes what goes in and what comes back.

What should you do next on your own entries?

Pull the "X" and "V" lines and every invoice description that says set, kit or bundle from the last year of entries. For the ten largest by duty, write down the three-part test answer and the essential-character component, with component costs and weights. Recompute each at the stack for its entry date both ways, and file the corrections the deadlines still allow. The post-entry audit pillar covers the other checks to run on the same lines.