Chinese USCC Codes and Supplier Document Checks

A Chinese supplier’s quotation, business licence and product label may show several names: a Chinese registered name, an English trading name and a brand. The unified social credit code, or USCC, gives you a separate identifier to record alongside those names. Checking its characters is a useful first step when reconciling documents, but it does not establish who produced the goods or whether the supplier can fulfil an order.

The code is an organizational identifier, not a numerical credit score. A large value is not a better rating, and passing a check-digit test is not a favourable credit assessment. ChinaClue checks the number’s structure under GB 32100-2015 and its published amendment; registration information must be checked separately. Keep the original document and its date so that a later correction remains traceable.

Separate the five encoded fields

The eighteen characters divide into five parts: one registration-department character, one organization-category character, six digits for the registration authority’s administrative area, nine characters identifying the organization and one final check character. Count positions rather than splitting wherever a letter appears. Letters are part of the identifier, not a prefix that can be discarded.

The tool’s worked example is 91350100M000100Y43, split as 9 | 1 | 350100 | M000100Y4 | 3. The first two characters encode the market-regulation registration department and enterprise category. The six-digit field is 350100; the organization-identifier field is M000100Y4. This example explains the calculation, not a recommendation or verification of a supplier.

Codes do not all begin with 91. Under the market-regulation department’s coding scheme, 92 identifies the individually owned business category and 93 the farmers’ professional cooperative category. Other registration departments have their own category combinations. A form that accepts only 91 would therefore exclude valid categories. Read these as encoded categories, not a live statement of the organization’s present form.

Worked example split into registration department 9, category 1, administrative code 350100, organization identifier M000100Y4 and final check 3. Lengths are 1, 1, 6, 9 and 1.

Read the two arithmetic checks separately

The outer code uses the 31-character alphabet 0123456789ABCDEFGHJKLMNPQRTUWXY. The letters I, O, Z, S, V are excluded. Character values come from positions in that alphabet, starting at zero; they are not ordinary alphabetical positions. For example, M has value 21 in this calculation. A letter can be legal in some other identifier and still be invalid here.

Multiply the first seventeen character values by these weights, from the left: 1, 3, 9, 27, 19, 26, 16, 17, 20, 29, 25, 13, 8, 24, 10, 30, 28. For the worked example, the weighted sum is 1640 and the remainder after division by 31 is 28. The check value is (31 − 28) mod 31 = 3, producing final character 3. A zero remainder produces zero, not an extra character for 31.

The nine-character organization identifier has its own final check character, calculated with a different alphabet and weights. In M000100Y4 that character is 4. ChinaClue reports this inner check separately from the full code’s final 3. A string can pass the outer calculation while failing the embedded organization check; one successful row must not hide another failed row.

Resolve transcription errors at the source

First compare the photographed or original licence with the copied string. ChinaClue normalizes fullwidth Latin letters and digits and accepts lowercase input by converting it to uppercase. Those presentation changes are different from deciding which character a blurred image contains. Keep the original value alongside any normalized version when preparing a document comparison.

Because O is excluded, OCR output containing O needs attention, but that does not prove every O should become 0. The source image may be incomplete, the wrong field may have been copied, or another digit may also be wrong. Re-read the relevant characters. Do not replace the final character with the calculator’s expected value merely to make the result green.

Store the code as text in a spreadsheet, preserving all eighteen characters. Do not let numeric import settings round an all-digit identifier or remove a leading character. Compare complete normalized codes, not only their last four characters. If two records have the same code but different names, preserve the discrepancy for investigation instead of silently merging the documents.

Keep historical codes distinct from current registration

The six-digit area field describes the registration authority’s administrative division in the coding structure. It is not the factory’s street address, a postal code or proof of where goods were manufactured. ChinaClue searches historical division records from 1981–2025. A retired code may have a historical match even when it is absent from the latest snapshot.

The management rules effective from 1 February 2026 specifically preserve the code when an individually owned business converts into an enterprise through change registration, and when a sole proprietorship converts into a company or partnership that way. Consequently, an encoded category alone may not describe the current registered form. Do not “correct” a retained 92 prefix to 91 by editing the identifier.

Use current registration records to establish the registered name, organizational form and status relevant to your task. Record when you checked them. A local arithmetic result contains no current registration response, cancellation history or business-operation assessment. A historical area match likewise cannot show whether an organization is still operating at an address today.

Build a document comparison that keeps roles visible

Make separate rows for the business licence, quotation, contract and product label. For each, retain the exact Chinese entity name, code if shown, document date and stated role. The party making a quotation and the manufacturer named on packaging can be different entities. Matching a brand name or English trading name is not a substitute for comparing the Chinese names and identifiers.

For example, a quotation may identify the seller while a tea package separately lists 委托方 and 受托方. Record those commissioning and commissioned parties separately; do not assign the seller’s code to every name on the package. A code’s final character does not encode a bank account, manufacturing permission or relationship between companies. Those are separate facts requiring their own evidence.

Finish with a precise finding: the entered characters passed or failed specified format and arithmetic checks; a historical code was found or not found; the documents show matching or differing recorded entities. Keep unresolved differences visible. This gives the next reviewer a usable record without turning a local format check into a claim that a business, transaction or product has been verified.