TariffWolf Logo TariffWolf

Reading the General Interpretative Rules Correctly

June 26, 2026 15 min read Blog
How are the WCO General Interpretative Rules correctly applied? This guide distils the right, ordered method from 20 classification disputes—GRI 1 to 6, essential character, and the parts rules.

An educational note on the disciplined application of the WCO General Rules for the Interpretation of the Harmonized System (Seventh Edition, 2022)


Why this matters

Almost every classification dispute that reaches a tribunal is, at its core, a disagreement about how the General Interpretative Rules (GRIs) are to be read — not about the facts of the goods. The product is usually agreed. What is contested is the method: which Rule applies, in what order, and what each Rule actually permits.

A study of twenty recent classification decisions makes this strikingly clear. Across vehicles, electronics, chemicals, metals, lamps, vessels and pharmaceuticals, the same handful of interpretive errors recur. They are not exotic. They are predictable, and therefore avoidable. This note distils the correct reading of the GRIs from those decisions, organised by the Rule most often misapplied.

The discipline can be stated in one sentence: the GRIs are a sequence, not a menu. They are applied in order; a later Rule is reached only when every earlier Rule has genuinely failed; and a Section or Chapter Note, where it speaks, overrides all of them.


First principles: the architecture the Rules assume

Three structural facts govern everything that follows.

1. Headings and Notes are paramount (GRI 1). Classification is determined first by the terms of the four-digit headings and the relative Section and Chapter Notes. Section, Chapter and sub-Chapter titles are provided “for ease of reference only”; they have no legal force. Goods are classified by what they are in the terms of the heading, read with the Notes — not by the chapter title, not by the importer’s commercial label, and not by intended end-use unless the heading itself makes use the test.

2. The Rules run in strict order. GRI 2 is reached only if GRI 1 leaves the matter open. GRI 3 operates only where goods are prima facie classifiable in two or more headings. Within GRI 3, the order of priority is fixed: (a) most specific description, then (b) essential character, then (c) the heading last in numerical order. GRI 4 (most akin) is a last resort, and GRI 5 (containers and packing) and GRI 6 (subheadings) sit on top of the rest. Skipping a step — or jumping to “essential character” before asking whether two headings genuinely compete — is the most common structural error in the case set.

3. Notes outrank Rules. The Explanatory Note to GRI 3 puts it plainly: the Rule “can only take effect provided the terms of headings or Section or Chapter Notes do not otherwise require.” Where a Note assigns or excludes goods, that Note governs and GRI 2 to 5 cannot displace it. Several disputes in the set were resolved entirely by a Chapter or Section Note, with the rival GRI arguments never reaching the point of decision.


Lesson 1 — GRI 1: classify the article by its own identity, not by use or by the chapter you would like to land in

The largest single source of error under GRI 1 is substituting end-use for the terms of the heading. A heading that names an article by what it is prevails over an argument that the article is used in something else.

Three patterns from the case set illustrate the discipline:

  • A structure is a structure, even when it is destined for a machine. Where shaped metal sections were imported to be built into agricultural equipment, the argument that they were therefore “parts of agricultural machinery” failed. The Section Notes exclude articles of base-metal structures from the machinery-parts route, and the goods answered the specific description of a structure of their constituent metal. End-use did not convert a structure into a machine part. The lesson: when an article has its own specific identity heading, you must classify it there before reaching for a “parts of X” heading justified only by where it will eventually be fitted.
  • Semi-manufactured material is not a finished article. Precious-metal chain imported in continuous running lengths was claimed as a finished article of jewellery. But the Chapter Notes distinguish semi-manufactured forms of the metal from finished articles. Material that has not yet become an identifiable finished article is classified as semi-manufacture. The deciding question under GRI 1 was not “what could this become?” but “what is it, as presented?”
  • A preparation is not a separate chemically defined compound. A product offered as a defined inorganic compound (and therefore a Chapter 28 good) was found to be a preparation/mixture falling outside Chapter 28 by virtue of the Chapter’s own Note restricting it to separate chemically defined elements and compounds. The Note, not a GRI, decided the case.

In each, the correct method was identical: read the heading text, read the controlling Note, and ask what the article is — not what it is for.


Lesson 2 — GRI 2(a): the “as presented” doctrine and the meaning of “essential character”

GRI 2(a) is the most litigated Rule in the entire set, and the most misunderstood. Its text rewards close reading:

Any reference in a heading to an article shall be taken to include a reference to that article incomplete or unfinished, provided that, as presented, the incomplete or unfinished article has the essential character of the complete or finished article. It shall also be taken to include a reference to that article … presented unassembled or disassembled.

Three principles flow from this, and the case set turns on all three.

2(a)(i) — “As presented” is assessed on the goods actually before the assessing officer, consignment by consignment

The Rule speaks of the article “as presented.” That phrase is doing real work. The essential-character test is applied to what is physically presented for assessment, in the state and quantity in which it arrives. It is not an invitation to gather up everything an importer has ever brought in, spread across many separate consignments and dates, and ask whether the aggregate would make a complete article.

A recurring error in the set was exactly this aggregation: goods imported under numerous separate entries, over months, were notionally combined and then reclassified as complete articles in unassembled form. The decisions rejected the manoeuvre. Each consignment is assessed on its own contents. You cannot manufacture “essential character” by summing shipments that were never presented together.

2(a)(ii) — “Essential character” requires the defining functional core to be present

Where the goods genuinely presented together lack the component that gives the finished article its defining function, GRI 2(a) does not apply. A vivid example from the set: kits of vehicle components imported without the battery — the power source being the defining element of an electric vehicle — were held not to possess the essential character of the complete vehicle. The absent core feature was decisive. “Essential character” is a finding, supported by reasons (the role of the missing or present component in the working of the article), not a label to be attached because most of the parts are present.

2(a)(iii) — But the Rule cuts both ways: a genuinely complete kit is the finished article

GRI 2(a) is not a one-directional shield for importers. Where the components presented together do amount to a complete article needing only assembly by fixing, bolting, riveting or welding — with no further working — the goods are the finished article, classified in its heading. In the set, genuinely complete knock-down kits of finished products (a lamp, a vehicle) were correctly classified as the complete article, not as “parts,” precisely because, as presented, they had its essential character and required only assembly.

The Explanatory Notes sharpen the boundary:

  • Blanks (articles already having the approximate shape of the finished article, usable only for completion into it) are covered by the Rule. Semi-manufactures that do not yet have the essential shape of the finished article — bars, discs, tubes — are not.
  • “Assembly” means assembly only. Components requiring further working operations to reach the finished state fall outside the Rule.
  • Excess components beyond the number needed for one complete article are classified separately.

2(a) codified: the vessel example

The same logic appears, hard-wired, in a Chapter Note. For ships and floating structures, the Note provides that a hull, or an unfinished or incomplete vessel, “is to be classified in heading 89.06 if it does not have the essential character of a vessel of a particular kind.” A dispute in the set over whether a bare hull had reached the essential character of a particular kind of pleasure craft was governed by exactly this test. The Note is GRI 2(a) made specific — and a reminder that the essential-character question is often answered by the Notes before the general Rule is even reached.

That same Chapter also excludes separately presented parts of vessels (other than hulls), which decided a further dispute: components of a dredging vessel brought in as separate entries could not be swept together as a single complete article, both because they were separately presented and because the Chapter excludes such separately presented parts.

The GRI 2(a) checklist that the cases reward:

  1. What is actually presented in this consignment? (Not across all consignments.)
  2. Does it have the essential character of the complete article — i.e., is the defining functional core present? State the reason.
  3. If unassembled, does completion require only assembly, or further working?
  4. Are there excess components? Classify them separately.

Lesson 3 — GRI 3: only when two or more headings genuinely compete, and then in strict order

GRI 3 is reached only where goods are prima facie classifiable in two or more headings. Two errors recur: invoking GRI 3 when in truth one heading already governs under GRI 1, and within GRI 3 jumping straight to “essential character” without first applying “most specific description.”

3(a) — Most specific description, and the “equally specific” trap

A heading that names the article, or identifies it more clearly and completely, is more specific than one describing it by a general class. But the Rule contains an important limit: where two headings each refer only to part of the materials in composite goods, or part of the items in a set, they are deemed equally specific — even if one reads more precisely — and the analysis must move to 3(b). Treating a part-only description as if it were a complete one is a frequent misstep.

3(b) — Essential character is a reasoned finding

GRI 3(b) applies only to four categories: mixtures; composite goods of different materials; composite goods of different components; and goods put up in sets for retail sale. The Explanatory Note lists the factors that may determine essential character — the nature, bulk, quantity, weight or value of a component, or the role of a constituent in relation to the use of the goods. Whenever 3(b) is invoked, the classifier must name the factor relied on. “Essential character” asserted without a stated basis is not a finding.

Multifunction and connected goods. Two disputes in the set concerned devices that do more than one thing — a wrist-worn device with communication functions, and personal audio devices with wireless connectivity. The discipline here is to identify the principal function: the feature that defines the article’s use. Connectivity that merely supports the primary function does not redefine the article. Classifying a device by a secondary or enabling feature, rather than by what it principally does, was the error in each. (Where a Section Note prescribes classification by principal function for composite machines, that Note governs the analysis directly.)

Sets put up for retail sale — the three-part test, applied strictly

A “set put up for retail sale” exists only where all three conditions are met:

  1. at least two different articles prima facie classifiable in different headings;
  2. put up together to meet a particular need or carry out a specific activity; and
  3. put up for sale directly to end users without repacking.

If any limb fails, the components are classified separately — the set rule does not apply. The disqualifying condition most often misread is the second: the items must be combined to meet one particular need or activity together. A bundle of unrelated articles does not become a “set” merely because it is sold in one package.

3(c) — Last in numerical order

Only when both 3(a) and 3(b) genuinely fail does 3(c) apply: classify in the heading occurring last in numerical order among those equally meriting consideration. It is a tie-breaker of last resort, not a shortcut to be reached for when essential character is merely difficult to articulate.


Lesson 4 — Parts, wholes and specific identity: the Section XVI Note 2 discipline

A distinct cluster of disputes turned not on the GRIs at all, but on the parts rules in the Section and Chapter Notes — which, being Notes, outrank the Rules. The governing principle (expressed in Section XVI Note 2, and mirrored for instruments in Chapter 90 Note 2) is a two-step hierarchy:

(a) Parts which are goods included in any of the headings of the relevant Chapters “are in all cases to be classified in their respective headings”; and only (b) other parts, if suitable for use solely or principally with a particular machine, are classified with that machine.

Three consequences, each tested in the set:

  • A good that has its own heading stays in its own heading, even when used as a part. An accumulator that answers the description of its own heading is classified there, not as a “part” of the appliance it powers — because limb (a) of the Note applies before limb (b). The “it is used inside the phone, therefore it is a part of the phone” argument fails at the first step.
  • Conversely, “used only in machine X” does not automatically send a good to machine X’s parts heading. Dedicated items that are themselves goods of a specific heading must be classified in that specific heading, not routed wholesale into a parts heading on the strength of their dedicated use. Limb (a) is checked before limb (b), every time.
  • GRI 2(a) cannot be used to convert a component into the finished article it feeds. Where an imported article is itself a complete, functional good of its own heading (for example, a working dispensing pump), it is classified as that good — not reclassified as a “part” of the larger finished article into which it will be incorporated, and not promoted to the finished article via GRI 2(a) unless, as presented, it has that article’s essential character.

The method is always the same: before defaulting to any “parts of …” heading, check the relevant Section/Chapter Note and ask whether the article is itself a good of a specific heading. If it is, that heading governs.


The recurring fault lines, summarised

Across the twenty disputes, the misinterpretations reduce to five:

  1. Aggregating separate consignments to build “essential character” under GRI 2(a), contrary to the “as presented” requirement.
  2. Reading “essential character” without the defining functional core present (or, conversely, refusing to recognise a genuinely complete knock-down kit as the finished article).
  3. Classifying by end-use into a “parts” heading when the article has its own specific identity heading.
  4. Classifying multifunction or connected goods by a secondary feature instead of by principal function.
  5. Reaching for a Chapter heading without first applying the Note that includes or excludes the goods — forgetting that Notes outrank the Rules.

A disciplined working sequence

For any classification, the case set rewards the following order of questions:

  1. Identify the goods as presented — composition, function, state (assembled, unassembled, incomplete), packaging, and what is in this consignment.
  2. GRI 1: Do the heading terms and the relative Section/Chapter Notes resolve it? Check every Note that includes or excludes the goods. If a Note governs, stop.
  3. GRI 2(a): Only if the goods are incomplete/unfinished or unassembled. Apply the “as presented” and essential-character tests, with stated reasons. Classify excess components separately.
  4. GRI 2(b) / GRI 3: Only if the goods are prima facie in two or more headings. Apply 3(a) → 3(b) → 3(c) in order, never skipping. Name the essential-character factor if 3(b) is used. Apply the three-part test strictly for sets.
  5. GRI 4 / 5: Most-akin as a true last resort; containers and packing per Rule 5.
  6. GRI 6: Descend to the six-digit subheading using Rules 1–5 mutatis mutandis, comparing only subheadings at the same dash level, and applying any Subheading Note.
  7. Stop at six digits. The Harmonized System is harmonised only to six digits; national breakdowns below that are a matter for the applicable national tariff schedule.

A closing observation on consequences

A theme that runs quietly through the decisions deserves mention. Where the dispute was a genuine, bona fide difference of opinion on classification — a real two-position question on which reasonable classifiers could differ — the tribunals were markedly reluctant to sustain confiscation, redemption fine or penalty. A defensible reading of an ambiguous heading is not the same as misdeclaration. This is not legal advice, and outcomes turn on the facts and the governing statute; but as a matter of classification practice it underlines why the reasoning matters as much as the result. A classification position built transparently on the correct GRI sequence, with the controlling Notes identified and the essential-character finding reasoned, is both more likely to be right and better protected if it is wrong.

The General Interpretative Rules are not a set of competing arguments to be selected among. They are an ordered method. Read in sequence, with the Notes given their paramount place and “essential character” treated as a finding rather than a slogan, they resolve the great majority of disputes before they ever begin.


Prepared as an educational reference. Classification reasoning is grounded in the WCO General Rules for the Interpretation of the Harmonized System and the Explanatory Notes, Seventh Edition (2022). Party names, adjudicating authorities and case citations have been omitted; the cases are used only to illustrate recurring interpretive principles. This note does not constitute a binding ruling or legal advice.

Share: