If you are searching for how to translate HS codes, how to translate HTS codes, how to translate Schedule B numbers, or how to handle customs commodity codes in multilingual import and export documents, the first rule is that the classification code itself is controlled data, not ordinary prose. Translate the product description, explanatory notes and surrounding customs language; preserve the authoritative code exactly unless a competent classification authority has determined that a different jurisdiction-specific code is required.
This distinction matters in commercial invoices, packing lists, customs declarations, product catalogues, tariff databases, freight documents, certificates, broker instructions, ERP systems, e-commerce exports and cross-border compliance workflows. A translation can be linguistically elegant yet operationally wrong if it silently changes a six-digit Harmonized System code, substitutes an import tariff number for an export Schedule B number, or assumes that a target-language product name automatically belongs in the same customs classification.
This guide explains how to translate HS, HTS, HTSUS, commodity and Schedule B information without changing classification meaning. It shows how the six-digit international HS foundation relates to national extensions, why edition and jurisdiction matter, how to translate official descriptions without turning translation into unauthorized customs classification, how to protect codes in spreadsheets and localization tools, and how to verify that a multilingual record still describes the same product under the same classification decision.
1. A customs code is a classification result, not a word-for-word translation
A tariff or commodity code expresses a legal and technical classification decision. The words beside it help humans understand the goods, but the number is not generated by translating those words into another language. Two products with similar everyday names can classify differently because of material, function, composition, processing, condition or the legal notes that govern a heading. Conversely, one code can cover products described with many different commercial names.
The translator’s task is therefore to preserve the relationship between the code and the product description. If the source supplies an approved code, that code is protected data. If the project requires classification for a different jurisdiction, that is a separate classification task that should be performed by a competent customs specialist, broker, ruling process or authorized internal function.
2. What the Harmonized System does
The Harmonized Commodity Description and Coding System, usually shortened to HS, is maintained internationally by the World Customs Organization. It organizes traded goods into a structured nomenclature and uses a common six-digit core. The structure supports customs tariffs, trade statistics and many other trade processes, but individual customs territories can extend the six-digit foundation for their own tariff or statistical needs.
That means “HS code” is sometimes used casually for several related things. A six-digit HS subheading is not automatically identical to a country’s eight-, ten- or other-digit tariff line. Translation teams should record exactly which code system, jurisdiction and edition the source uses.
3. Read the hierarchy before you translate the description
At a simplified structural level, the first two digits identify an HS chapter, four digits identify a heading and six digits identify a subheading. The hierarchy matters because a description at one level may be broader than the description at the next. A translator should not expand a short heading label into a narrower product claim unless the source context justifies it.
The wording of tariff nomenclatures is compact because it works together with section notes, chapter notes, subheading notes and interpretive rules. Translating the label without the surrounding legal structure can make a phrase sound more definite than the classification system intends.
4. Six digits are international; extensions are jurisdiction-specific
The international HS foundation gives trading partners a common starting point through six digits. National or customs-territory schedules can add digits after that point. Those extra digits are not decorative precision. They can carry tariff, statistical, quota, regulatory or reporting distinctions that exist only in that jurisdiction.
A target-language document for another market should not automatically replace a source national extension with a locally familiar one. First determine the task: are you translating the same declaration, preparing a new import entry, preparing an export filing or creating a multilingual catalogue? The correct code treatment depends on that operational purpose.
5. HTS and HTSUS are not generic synonyms for every tariff schedule
In United States import work, HTS or HTSUS commonly refers to the Harmonized Tariff Schedule of the United States. It builds on the HS foundation and contains U.S.-specific tariff and statistical detail. A translator handling a U.S. import document should preserve the actual HTSUS code and current schedule context rather than replacing it with a bare six-digit HS number because that looks more “international.”
Likewise, a code from another customs territory should not be relabeled “HTSUS” merely because the target language is English. Language and jurisdiction are separate axes. English can describe a French, Singaporean, Japanese or Brazilian tariff code without converting that code into a U.S. tariff line.
6. Schedule B is an export classification system, not simply another name for HTSUS
The U.S. Schedule B is used for export statistical classification. Schedule B numbers are ten-digit numeric codes. They are related to the HS structure but are not universally interchangeable with U.S. import HTSUS numbers at the full national-extension level. A multilingual workflow should keep the fields named accurately so exporters do not mistake an import classification for an export reporting code.
If a source system contains both HTS and Schedule B fields, preserve both. Do not deduplicate them because their first six digits happen to match. Their downstream uses are different.
7. Edition and revision date are part of classification meaning
The HS is revised periodically, and national tariff schedules can change more frequently. A code that was valid in an earlier edition can be split, merged, renumbered or reworded later. Translation memory can therefore preserve obsolete classification language long after the code has changed.
Every production workflow should know the effective date or edition that governs the source. When a translation is archival, preserve the historical classification. When the target will be used for a new transaction, confirm the current code with the responsible classification function rather than assuming the old number remains valid.
8. Correlation tables help navigation but do not replace classification
When nomenclatures change, authorities may publish correlation or concordance tables linking old and new provisions. These tables are valuable for research, but a one-to-many or many-to-one change often means the old description does not contain enough information to choose the new code automatically.
A translation team can use concordances to flag records needing review. It should not silently migrate every historical code on the strength of a spreadsheet lookup unless the classification owner has approved that process.
9. Translate the legal description at the right level of precision
Tariff descriptions use controlled nouns, qualifiers, exclusions and residual phrases such as “other.” The target text should preserve scope. Adding a familiar commercial term may narrow the provision; omitting a modifier may broaden it. Punctuation can also signal how qualifiers attach to a list.
The safest method is concept-first translation: identify the product class, material, function and grammatical relationships before choosing target terminology. Do not polish away awkward-looking legal structure if that structure carries classification boundaries.
10. Commercial product names are evidence, not classification rules
Marketing names are optimized for customers, not customs nomenclature. A “smart bottle,” “travel organizer” or “gaming accessory” may require a classification analysis based on physical characteristics and function. Translators should preserve commercial branding accurately while keeping it distinct from the customs description.
If a client asks the translator to “translate the product name into the HS code,” that is not a translation operation. It is a classification request. Route it accordingly.
11. Material can control classification
Textiles, plastics, metals, paper, wood, ceramics and composite materials often require precise material descriptions. “Metal part” may be insufficient when the source distinguishes steel from aluminum or a coated material from an uncoated one. Translation that generalizes the material can deprive a classifier of evidence needed for the code.
Glossaries should therefore include material vocabulary, production state and composition terms. These are not merely stylistic choices; they can affect classification reasoning.
12. Composition percentages must remain attached to the right substance
Food, chemicals, textiles and mixed materials may be described by percentages. Translators must preserve the number, unit and material relationship. A percentage moved to the wrong ingredient or fiber can change how a product is understood even if every individual token is copied correctly.
For structured data, keep composition in separate fields when possible. For prose, use grammar that makes the attachment unmistakable.
13. Function matters, but intended use is not a magic label
Machinery, electronics, instruments and parts are often described by what they do. Translating function accurately helps the classification specialist, but the everyday purpose stated in marketing material may not be the only legally relevant criterion. Notes and heading text can control.
Avoid strengthening a source phrase such as “suitable for” into “designed exclusively for.” That change can create false classification evidence.
14. Parts and accessories require controlled terminology
The words “part,” “component,” “accessory,” “attachment” and “spare” are often used loosely in business English. Customs nomenclature can treat these concepts differently. Translate the source term in context and preserve technical relationships such as dedicated use, general use, constituent material and whether the item is separately functional.
If the source itself is vague, the translation should not invent legal precision. Flag the ambiguity for product or classification experts.
15. Sets and kits are not solved by translating the box label
Retail sets, repair kits and multi-component packages can contain goods that would classify differently if imported separately. A translator should accurately describe each constituent item, quantities and packaging rather than compressing the target into a catchy set name.
Classification of sets can depend on interpretive rules and essential-character analysis. That decision belongs outside pure translation.
16. Unfinished and incomplete goods need state-of-completion language
Words such as blank, unfinished, unassembled, semi-finished, knock-down, pre-cut and machined can be decisive. A target that simply says “part” can erase the product state the classifier needs.
Preserve whether components are assembled, packaged together, processed to a particular stage or merely intended for later manufacture. Translation should expose the evidence, not collapse it.
17. Packaging is different from the goods it contains
Cartons, bottles, cases, cylinders and reusable transport equipment may appear in the same record as the product. A code for the goods should not migrate to a packaging row, and the packaging description should not be merged into the commodity name.
When packaging itself is traded or separately accounted for, its classification can become a distinct question. Maintain row and field boundaries.
18. Country of origin and HS code answer different questions
Origin rules determine the economic nationality of goods for particular trade purposes. HS classification describes what the goods are under the nomenclature. A multilingual invoice may contain both. Translators should not infer one from the other or translate “origin” as if it meant shipment country, manufacturing site or tariff code.
The same classification can apply to goods of different origins, and origin rules can depend on processing history. Keep the concepts separate.
19. Tariff rate is not embedded forever in the code
A classification code can lead to a tariff treatment in a given jurisdiction and date, but duties can also depend on origin, trade programs, quotas, remedies and other measures. Translators should not append a duty percentage to a code unless it comes from the authoritative source for the relevant transaction.
A product catalogue that preserves a code for reference should avoid implying that one universal duty rate follows the product worldwide.
20. Export-control classifications are separate systems
Export-control identifiers such as ECCNs serve different regulatory purposes from HS or Schedule B codes. A product may have both. The two should not be merged into a generic “commodity code” field during translation.
If a source uses acronyms ambiguously, resolve the field through the system schema or compliance owner rather than guessing from the number length.
21. Dangerous-goods numbers are also separate
UN dangerous-goods numbers identify regulated substances or articles for transport. They are not HS classification codes. A chemical shipment may contain CAS identifiers, UN transport numbers and customs codes together. Each system answers a different question.
Translate the explanatory language around these codes, but keep the identifiers in their own fields and preserve cross-references accurately.
22. A reliable workflow: map every classification-bearing field before translation
Step 1: inventory code systems
Identify HS, HTSUS, Schedule B, local tariff, export-control, dangerous-goods and internal product-classification fields. Name each system explicitly.
Step 2: record jurisdiction and edition
For national tariff data, capture the customs territory and effective schedule. For historical documents, preserve the relevant historical context.
Step 3: protect numeric strings
Store codes as text so leading zeros and fixed widths survive. Lock them from translation, numeral localization and spreadsheet reformatting.
Step 4: translate product evidence
Translate composition, material, function, processing state, dimensions, technical specifications and commercial description with controlled terminology.
Step 5: separate translation from classification decisions
If the source supplies a code, preserve it. If a new code is required, route the record to the appropriate classification authority.
Step 6: validate against authoritative schedules
Confirm that the code exists in the stated schedule and edition. A valid-looking number is not enough.
Step 7: compare source and target code sets
Use automated QA to detect added, removed or mutated codes.
Step 8: verify row relationships
Ensure each classification remains attached to the same SKU, product description, quantity and jurisdictional field.
Step 9: review exceptions
Flag discontinued codes, invalid lengths, mismatched descriptions and ambiguous system labels for human resolution.
Step 10: retain an audit trail
Record who supplied or approved classifications and when the schedule was checked. This separates linguistic accountability from classification accountability.
23. Spreadsheets can silently convert customs codes into the wrong data type
A numeric-looking classification may lose leading zeros, appear in scientific notation or be rounded if software treats it as a number. That is a data error, not a translation issue, but it often enters through translation handoffs.
Import classification fields explicitly as text. Define expected lengths and validate them after export. Never rely on visual inspection of a formatted spreadsheet alone.
24. Punctuation should follow the governing display convention
Some systems show tariff numbers with periods or spaces; others store a continuous digit string. Presentation punctuation is not the same as the underlying classification value. Define whether the project preserves source formatting or converts to a canonical system format.
Do not invent punctuation according to target-language number style. A thousands separator inside a tariff number can create a nonexistent code.
25. OCR can confuse codes with invoice values
Scanned invoices often place HS codes next to quantities, prices and line numbers. OCR can merge columns, drop digits or assign a code to the wrong item. Before translation, validate table structure and expected code length.
When possible, obtain the original electronic export from the ERP or customs system. Structured data is safer than reconstructing classification from an image.
26. Translation memory should remember descriptions, not stale classifications
A previous segment may contain a product description plus last year’s tariff number. A 100 percent translation-memory match can therefore be linguistically perfect and operationally obsolete. Treat tariff codes as variables and force them from the current source.
Terminology memory and classification data should also have different ownership. Linguists can maintain approved equivalents; classification specialists maintain the legally operative code decision.
27. Machine translation should receive protected code tokens
Translation engines generally do not need to process a numeric tariff code. Mask or lock the field. The engine should work on descriptions, notes and instructions while the code passes through unchanged.
After translation, restore protected values from the source data layer and compare exact code sets. This makes QA deterministic.
28. Generative AI must not classify by confidence-sounding guess
A language model can propose plausible HS headings from a product name, but plausibility is not an authoritative classification. Missing details about material, composition, function or legal notes can change the result. Do not present model output as a customs determination.
AI can help organize product facts, detect inconsistent descriptions or explain terminology, but production classification should follow the responsible legal and operational process.
29. Right-to-left layouts need code isolation
Arabic and Hebrew documents can visually reorder digits and punctuation if bidirectional handling is poor. Keep classification fields as isolated left-to-right data where the receiving system expects Latin digits.
Test copying and searching from the final PDF or application. A code that looks correct but copies in a different order is not safe.
30. Do not localize digits inside authoritative code fields
A target locale may display another numeral glyph set in prose. Customs systems frequently expect ASCII digits in code fields. Preserve the machine-facing representation unless the receiving authority explicitly specifies otherwise.
If a human-readable document uses localized numerals, maintain a separate canonical value for electronic filing and reconciliation.
31. Product dictionaries need concept definitions, not only bilingual pairs
A glossary entry for “upper,” “liner,” “substrate,” “assembly,” “concentrate,” “prepared,” “coated” or “woven” should explain the concept and domain. This helps translators preserve distinctions that may matter to classification.
The protected Vocabulary Learning Hub is conceptually relevant here because accurate word choice supports accurate product evidence. It does not own the customs classification decision; this article remains responsible for the code-and-description translation problem.
32. Grammar can change classification evidence
Compare “plastic housing for a pump” with “pump in a plastic housing.” The same nouns appear, but the grammatical head changes what object the phrase describes. Technical noun strings, participles and prepositional attachments require careful parsing.
This is where eduKateSG’s How English Works ecosystem supports the translator: grammar helps determine which property belongs to which object before terminology is chosen.
33. Made of, containing and coated with are not interchangeable
Material relations can affect product interpretation. A steel article coated with plastic is not the same description as an article made of plastic containing steel. Translators should preserve the source relation rather than reducing every material phrase to a noun list.
For multilingual product master data, structured material fields often reduce ambiguity better than free-text descriptions.
34. Measurements must stay attached to the correct feature
Thickness, diameter, capacity, wattage, weight and concentration can distinguish product variants. If the target moves a measurement from the core product to its packaging or accessory, the classification evidence changes.
Treat number-unit-property triples as a semantic unit during translation and QA.
35. Model numbers and SKUs are not customs codes
A manufacturer model number or internal SKU identifies a commercial item. The customs code classifies it under a nomenclature. One tariff classification may cover many SKUs, and one SKU may require different national extensions across markets.
Preserve both. Do not substitute the SKU into a missing tariff field or vice versa.
36. A multilingual catalogue needs jurisdiction-specific classification columns
Global product catalogues often fail when they provide one column called “HS/HTS code” and assume it applies everywhere. A stronger design separates six-digit HS reference, U.S. HTSUS, U.S. Schedule B and other market-specific codes as appropriate to the business.
Translation then localizes descriptions while the classification layer is maintained by jurisdiction. This reduces accidental reuse of one country’s extension in another market.
37. Worked example: a six-digit HS reference plus a national extension
A source catalogue contains a six-digit HS subheading for worldwide reference and a ten-digit national import line for one market. The French target translates the product description but preserves both numeric fields and their labels. It does not convert the national line merely because the interface language changed.
Language changed; customs territory did not. That distinction prevents needless recoding.
38. Worked example: HTSUS versus Schedule B
An American company imports a component and later exports a finished product. Its system stores an HTSUS import code for the component and a Schedule B export code for the finished product. A translator preparing multilingual documentation must not copy one code into the other field simply because both belong to U.S. trade data.
The underlying transactions, products and classification systems are different. Field labels and workflow context must survive translation.
39. Worked example: a historical code
An archive contains a customs declaration from several years ago using a code that has since been replaced. The translation is evidentiary, so it reproduces the historical code and translates the historical description. A note can identify the schedule year if needed.
Updating the code to today’s schedule would rewrite the historical record rather than translate it.
40. Worked example: product description too vague to classify
A source says only “adapter.” The translator can render that word, but no responsible workflow should infer a customs classification from it. Is it an electrical adapter, pipe fitting, optical adapter, mechanical coupling or something else? The source lacks necessary product facts.
The correct action is to request technical detail. Translation cannot manufacture evidence that the source never contained.
41. Worked example: a model proposes a plausible code
An AI system suggests an HS heading from a web-shop title. The company treats the suggestion as a research lead, not a filing decision. A specialist checks composition, function, notes and current tariff schedule before assigning the operational code.
This separation lets AI assist without disguising uncertainty as legal certainty.
42. Worked example: a row-shift error
A translator sorts product descriptions alphabetically but forgets to include the protected code column in the sort range. Every code remains syntactically valid, yet all codes now describe the wrong products. Exact-string QA passes and classification integrity fails.
Use stable row keys and relational comparisons. The central question is not merely “Did the code survive?” but “Did the code stay with the same product?”
43. Quality assurance should test four dimensions
Syntax: does the code have the expected character pattern and length? Authority: does it exist in the stated schedule and edition? Relationship: is it still attached to the same product and jurisdiction? Language: does the target description preserve the factual attributes relevant to the source? All four matter.
A workflow that checks only spelling misses classification data; a workflow that checks only numbers misses semantic drift in the descriptions.
44. When to escalate
Escalate when a code is invalid in the stated schedule, the source and product facts conflict, a requested target jurisdiction needs a new national extension, an old code has split into several new provisions, the product description lacks material or function detail, or a client asks the translator to make a classification decision beyond the agreed scope.
Escalation is a sign of control, not failure. Customs classification depends on evidence and jurisdiction.
45. What translate HS code usually means in search
People searching this phrase may want to translate the official description, understand what a code means, map an HS code to a national tariff schedule, find an export code, translate a customs invoice or verify that multilingual product data still carries the right classification. These are different intents.
A useful answer begins by identifying the code system and jurisdiction. Only then should language work begin.
46. What not to infer from a tariff code
A tariff code does not by itself prove origin, duty rate in every market, export-control status, dangerous-goods status, product safety compliance, valuation or admissibility. Each of those has its own rules and evidence.
Translate only the claim the source actually makes. Do not turn a classification number into a general certificate of regulatory compliance.
47. Reference route for international HS verification
For the international system, use the World Customs Organization’s Harmonized System overview and the applicable nomenclature edition. Because revisions occur, verify the edition effective for the transaction date instead of relying on an undated third-party code list.
When a classification issue is material, consult the responsible customs authority, broker, binding-ruling process or qualified specialist for the jurisdiction concerned.
48. Reference route for U.S. import and export codes
For United States imports, use the current Harmonized Tariff Schedule maintained by the U.S. International Trade Commission. For export Schedule B information, use the U.S. Census Bureau Schedule B resources. These sources serve different operational purposes and should remain labeled accordingly in multilingual systems.
Do not let a translation-memory match or search-engine snippet outrank the current authoritative schedule.
49. Connection to the translation architecture
This article belongs under eduKateSG’s Master Art of Translation — Technical Translation System. That owner explains the broader system of specifications, terminology, standards, QA and change control. This page has a narrower job: preserve customs classification identity while product meaning moves between languages.
Keeping those scopes separate prevents a specialist customs article from becoming a competing general translation hub.
50. Release checklist
Before release, confirm the code system, jurisdiction and edition; preserve every authoritative digit; protect leading zeros; distinguish six-digit HS from national extensions; keep HTSUS and Schedule B fields separate; verify product-code relationships; preserve material, composition, function and processing details; check units and percentages; inspect OCR and spreadsheet transformations; compare source and target code sets; and route unresolved classification questions to the proper authority.
Then perform ordinary linguistic QA for terminology, grammar, clarity and target-language usability. Both layers are necessary.
51. Change control belongs in the product master, not in ad hoc translation notes
When a classification changes, update it through a governed product-master or compliance process that records the prior value, new value, effective date, jurisdiction, reason and approver. Translators should consume that controlled data rather than maintaining private code corrections in local spreadsheets. Otherwise two language teams can end up publishing different classifications for the same SKU.
A translation note is useful for explaining an anomaly, but it should not become the hidden system of record. Classification needs one accountable owner and a visible revision history.
52. Country-specific names can conceal the same product and the same code
One product may be sold under different brand names or marketing descriptions in different countries. That does not automatically mean its six-digit HS classification changes. Conversely, keeping the same brand name does not guarantee the national tariff extension is the same everywhere. Translation teams should therefore key classification data to stable product identity and jurisdiction, not to localized marketing copy.
This prevents a common catalogue mistake in which a renamed product is treated as newly classified or a familiar brand name causes an old code to be reused without review.
53. Product variants should not inherit classification blindly
A family of products can differ by material, power source, composition, capacity, construction or included accessories. Copying the parent product’s code into every translated child SKU may be wrong. The translation workflow should preserve the classification supplied for each variant and flag empty fields rather than auto-filling them from a neighboring row.
Inheritance rules belong in master-data governance. Language systems should not create them accidentally through translation memory or repeated templates.
54. Supplier codes are inputs, not unquestionable authority
Suppliers often provide HS or tariff codes, and those codes can be valuable evidence. Importers, exporters or declarants may nevertheless have their own legal responsibilities for the classifications they use. A translator should preserve the supplier-provided value accurately while keeping its provenance visible.
If internal compliance later approves a different code, the change should come from that controlled process. The translator should not resolve a supplier-company disagreement by choosing the code that appears more plausible.
55. Multilingual search should index descriptions without rewriting classifications
A useful global catalogue can let users search translated descriptions, synonyms and product vocabulary while storing classification codes as stable facets or attributes. This improves discoverability without forcing the code itself to become language-dependent. The same six-digit reference can be found from many target-language queries.
Search engineering and customs classification should reinforce each other but remain distinct: search helps people find the record; the authoritative classification field tells them which controlled code the record currently carries.
56. Final rule: translate the description, preserve the classification decision
A customs code can travel through many languages without becoming a different code. What changes is the human explanation around it. The safest multilingual trade workflow lets language specialists make descriptions clear and accurate while classification specialists own the legal-technical decision. That division of responsibility keeps products understandable, codes traceable and declarations auditable across borders.