GRI 1 says classification is determined by the terms of the headings and any relative section or chapter notes. That is the whole rule, and it settles the large majority of classifications, because most products fit a heading by their description and the notes confirm it. The section and chapter notes are law, not guidance: a note that excludes a product from a chapter overrides the heading text, and a note that defines a term controls how that term is read. The classifier who skips the notes is the classifier who gets it wrong.
The practical habit is to read the heading, then read the notes for the section and the chapter, then decide. A product that looks like it belongs in Chapter 84 may be excluded by a note that sends it to Chapter 90. A product that looks like a part may be a good in its own right under a note that says so. The notes are short, they are binding, and they are where most classification errors hide.
GRI 2 covers the products that are not yet what they will be. Rule 2(a) says an incomplete or unfinished article is classified as the complete article if it has the essential character of the complete article, and an unassembled article is classified as the assembled article. Rule 2(b) says a mixture or combination of materials is classified by the component that gives it its essential character, or, if that cannot be determined, by the last rule that applies.
GRI 3 covers the products that fit more than one heading. Rule 3(a) prefers the most specific heading. Rule 3(b) prefers the heading for the component that gives the product its essential character. Rule 3(c) is the tiebreaker: when neither rule settles it, classify under the heading that occurs last in numerical order. GRI 4 is the fallback for products that fit no heading at all: classify under the heading for the goods to which they are most akin. GRI 5 covers containers and packing materials, which are classified with the goods they contain when they are of a kind normally sold with the goods. GRI 6 applies the same logic at the subheading level, comparing only subheadings at the same level.
The six rules are applied in order, and the order is not a suggestion. You do not jump to GRI 3 because GRI 1 is inconvenient. You do not invoke GRI 4 because GRI 3 gives you an answer you do not like. The sequence is the discipline: GRI 1 first, then 2, then 3, then 4, then 5, then 6, and you stop at the first rule that settles the question. A classification that cites GRI 3 without explaining why GRI 1 and 2 did not apply is an incomplete classification, and it will not survive a challenge.
The Harmonized System itself is large: 21 sections, chapters 1 through 97 with chapter 77 reserved, 1,228 headings, and 5,612 subheadings in the 2022 edition. The next edition, HS 2028, comes into force on January 1, 2028. The rules do not change with the edition. The headings do, which is why the classification has to be re-checked when the edition changes, even when the product has not. For the section where the rules get tested hardest, classifying a smartwatch, 8517 or 9102 walks a real dispute through the rules, and classifying a coffee maker shows how the notes decide a case the heading text leaves open.
The six rules, GRI 1 through 6, that decide how every product is classified under the Harmonized System. They are applied in order, and you stop at the first rule that settles the question.
GRI 1. It says classification is determined by the terms of the headings and the section and chapter notes, and it settles the large majority of classifications. The other rules exist for the cases GRI 1 cannot settle.
It is the tiebreaker. When a product fits more than one heading and neither the most-specific rule nor the essential-character rule settles it, classify under the heading that occurs last in numerical order.
January 1, 2028. The General Rules do not change with the edition, but the headings do, so classifications have to be re-checked when the edition changes, even when the product has not.