Browse the complete Harmonized System nomenclature, chapter by chapter, and jump straight from a classification to the live shipment records filed under it.
| Level | Digits | Example | Meaning |
|---|---|---|---|
| Chapter | 2 | 09 | Coffee, tea, mate and spices |
| Heading | 4 | 0901 | Coffee, whether or not roasted |
| Subheading | 6 | 090111 | Coffee, not roasted, not decaffeinated |
| National tariff line | 8–10 | 09011111 | Country-specific extension |
Every chapter links to its shipment records, top traders and price benchmarks.
The Harmonized System code is the international standard for classifying traded goods, maintained by the World Customs Organization. The first six digits are identical worldwide; countries add further digits for their own tariffs.
Two digits give the chapter, four give the heading, six give the subheading. India extends this to eight digits as ITC(HS); the US uses ten-digit HTS codes.
Start from the chapter that covers the material or function, then narrow through the headings. Cross-check against real shipment descriptions in the trade data — that shows the classification customs actually accepted.
Misclassification can mean incorrect duty, delayed clearance, penalties, or loss of preferential tariff treatment under a trade agreement.
The first six digits are. Beyond that each country maintains its own extensions, so an eight-digit Indian code will not match a ten-digit US code.
The WCO revises the nomenclature roughly every five years. National schedules are updated more frequently through budget notifications.
The 98 chapters are not an arbitrary list; they are grouped into sections that run broadly from the natural to the manufactured. Live animals and food occupy the early chapters, minerals and chemicals the twenties and thirties, plastics and leather the late thirties and forties, textiles the fifties and sixties, metals the seventies and eighties, machinery and vehicles the eighties, and instruments, furniture and miscellaneous articles the nineties. Knowing that shape means you can usually get to the right neighbourhood before you open the schedule at all.
| Chapters | Broad grouping | Examples |
|---|---|---|
| 01–24 | Agriculture, food and beverages | Live animals, cereals, spices, processed food |
| 25–27 | Minerals and mineral fuels | Salt, ores, petroleum and coal |
| 28–38 | Chemicals and allied industries | Organic chemicals, pharma, dyes, cosmetics |
| 39–43 | Plastics, rubber, leather and furs | Polymers, tyres, hides, leather goods |
| 44–49 | Wood, pulp and paper | Timber, plywood, paper, printed matter |
| 50–63 | Textiles and apparel | Silk, cotton, man-made fibres, garments, made-ups |
| 64–71 | Footwear, stone, ceramics and gems | Footwear, glass, ceramics, jewellery |
| 72–83 | Base metals and articles | Iron and steel, aluminium, copper, tools |
| 84–85 | Machinery and electricals | Mechanical appliances, electrical equipment |
| 86–89 | Transport equipment | Rail, vehicles, aircraft, ships |
| 90–98 | Instruments and miscellaneous | Optical, clocks, arms, furniture, toys, art |
Classification runs from the general to the specific and follows six interpretative rules that apply in order. In practice the sequence is: identify the plausible chapter from the grouping above; read the section and chapter notes, which are binding and frequently exclude what you expected; find the heading whose text most specifically covers the goods as presented; narrow to the six-digit subheading; then take it to the national tariff line. Most errors happen at the second step, because the notes are the part people skip.
The most reliable evidence for a classification is not an opinion about the tariff text — it is what comparable goods actually cleared under. Search the shipment record for the descriptions your industry uses, read the codes those consignments were filed against, and you have an administration’s own accepted answer for goods like yours. Where the duty difference is material and the position is genuinely uncertain, an advance ruling converts an open exposure into a settled one for the cost of some paperwork.
Duty is assessed at the national line — eight digits in India, ten in the United States. A six-digit code identifies the product internationally but cannot support an assessment, and a foreign supplier's full code belongs to their tariff, not yours.
Trade data rewards a short, boring discipline far more than it rewards technique. Four steps cover most of it, and skipping any one of them is where the confident wrong conclusions come from. Fix the classification first, because every filter, every duty figure and every price comparison downstream is keyed to the tariff line and inherits any error in it. Then read at least three years, so that seasonality and trend can be told apart rather than conflated. Then exclude the most recent one or two periods, which are still filling in as late filings arrive. Then separate value from volume, because a value movement can be price, quantity or a change of mix inside the line, and those point in different directions.
| Step | What it prevents | Cost of skipping it |
|---|---|---|
| Confirm the tariff line | Filtering the wrong product | Every downstream figure is wrong by an unknown amount |
| Read three years | Mistaking a season for a trend | Strategy built on a cyclical high or low |
| Drop the incomplete tail | Reading reporting lag as decline | Writing off buyers who never stopped buying |
| Separate value from volume | Reading price as demand | Investing against a movement that was not demand |
| Check the counterparty | Acting on an unverified name | Credit or capacity committed to a company with no history |
| Record the period and source | Unrepeatable analysis | Figures nobody can reconcile three months later |
The material here is one layer of a set that is meant to be used together. The country pages establish the shape of a market from official reported figures. The HS chapter pages take a single classification down to product level. The industry hubs group the chapters that make up a real industry and sum them, because almost no industry is one chapter. The India location pages read the national record as places, using the clearance point as a geographic signal. And the trade role pages take one job at a time — building an importer list, checking an exporter, reading a lane — and set out the signals that matter for it.
Reported totals, partner markets and chapter breakdown for 99 markets.
Learn more ›All 98 chapters, each with the markets that trade it.
Learn more ›Thirty-eight industries, each summed across the chapters it spans.
Learn more ›Seventy-nine trading places, their gateways and their clusters.
Learn more ›The same record read as a buyer list, a supplier check or a lane analysis.
Learn more ›Classification, documentation, pricing, sourcing and compliance in practice.
Learn more ›Customs records cover goods that physically crossed a border and were declared to an authority. They do not cover services. They do not cover domestic trade, so a business selling mainly inside its own market will look far smaller here than it is. They carry no margin, no contract terms, no payment behaviour and no intent. Coverage of counterparty names varies by jurisdiction and is not universal, data arrives on a lag, and published periods are revised as corrections come in.
None of that reduces what the record is good for, and stating it plainly is what makes the rest credible. Used within its limits, customs data is one of the very few commercial sources where the underlying event actually happened, was documented at the time, and was documented under legal obligation rather than for promotional purposes. That is a rare property, and it is worth not overselling.
Before relying on any trade dataset — ours or anyone else's — ask which markets are covered at which depth, what the lag is in each, how company names were matched, and what happens to a historical series across an HS revision. The answers tell you more than any headline figure.