Building Fitment Data for Pre-1981 Classics When the Databases Stop

NHTSA tells this story itself. In the background section of its 2008 VIN rulemaking the agency records that American manufacturers have used a vehicle identification number since 1954, and that the early VINs came in a wide array of configurations and variations depending on the individual manufacturer. Standard No. 115, effective January 1969, required a passenger car's VIN to be permanently marked and visible through the glazing — it standardised where you look, not what you find. The structured system arrived with a final rule published on 17 August 1978 (43 FR 36448) which, in the agency's own summary, stipulated that beginning with the 1981 model year all over-the-road vehicles sold must carry a 17-character VIN in a fixed format. Before that, the number is the manufacturer's own scheme, of its own length, with position meanings it chose — so the identifier that resolves a modern fitment question in one lookup resolves nothing here without marque knowledge you have to acquire and maintain.

The catalog layers stop in the same neighbourhood. ALLDATA describes its repair database as covering 1982 forward and routes pre-1982 vehicles to a Library Request its library team researches by hand — a good service, but a person answering a query rather than a database you can work at catalog scale. The fitment data is subtler, and is where generic articles mislead: the Auto Care Association's VCdb page advertises 175,500+ vehicle applications and 2.4+ million vehicle system configurations, and the 2026 VCdb Coverage chart it links shows U.S. light-duty coverage from 1896 to 2027. The year rows exist. But a coverage chart is a chart of year ranges, and makes no claim about how many engine, transmission or sub-model attributes hang off a 1968 row versus a 2018 one. Open the data for your own vehicles and count.

Not one source. Four, and the interesting information is where they disagree.

The backbone is the factory master parts catalog for the marque and year span, and its architecture is consistent enough to describe generically. The publicly posted demo of the 1965–72 Ford car master parts and accessory catalog is a good specimen: two volumes used together, one carrying numbers and one exploded illustrations; two major groups, chassis and body; each split into sections by category, each section running in sequential group-number order. Body items carry an extra two-digit body-type code. The prefix encodes decade, year of introduction or revision, model application and the design engineering office that owned the part, and a trailing suffix letter marks each successive revision — a running-change log hiding in plain sight.

Two warnings in that same demo belong on your parts-room wall. The lookup procedure assumes you already know the vehicle's year and car-line code, and sends you to the car's own warranty plate for them: the factory's answer to "which car is this?" is a tag on the vehicle, not a number in a book. And it states that engineering or casting numbers marked into a part during manufacture are not interchangeable with service part numbers, and that a service part may differ from the part installed on the line. That is the commonest source of bad application data in this business — somebody reads a number off a casting, calls it an application, and a hundred listings inherit the mistake.

The application listing is customer-facing: which years, models, body styles, positions and options the part covers and — the half people leave out — which it explicitly does not, with a version and a date because it will change. The manufacturing spec is what a factory works from: geometry, material, finish, tolerance, process, and no application data at all. The listing is supposed to change as you learn; the spec must never change quietly, because a changed spec is a different part and a different part needs a new first article.

That separation is why we describe parts to factories by vehicle make, model, year range and part type plus a drawing and a sample, and do not hand original-equipment reference numbers to offshore factories. It is a business decision rather than squeamishness: a reference number is a searchable key straight to whoever already produces the part, and it carries no manufacturing information the drawing does not.

Everything physical, usually better than you expect: wall thickness, draft, radii, weld and fastener patterns, spring rate, plating stack, material family subject to test. A good factory tells you what has to change to make the thing manufacturable — where the gate and ejector pins land, which radius is too tight for the press — and those conversations are the value.

What they cannot tell you is which model years your part covers, whether your donor is the early or the late variant, whether the mounting hole moved in mid-1967, or whether two millimetres of bow is design intent or fifty-five years of sag. The factory reproduces, faithfully, exactly the donor you sent, quirks and warp included; that is what reproduction from a sample means anywhere. So mark the sample up before it ships — what to reproduce, what to correct, which features must be measured to the drawing — and have the first article inspected against the drawing rather than the donor.

Treat a return as data, not just a refund. Capture the vehicle's year, body style and build date off its plate, what fouled and by how much, and a photograph with a rule in the frame. Then decide which of three things you have: a fitment-record correction, a variant split you did not know existed, or a tooling problem. Refunding without recording is how a business sells the same wrong listing for three years.

When you do find a mid-year change two years in, version the listing rather than overwrite it: split one record into two with a documented break point, keep the old one visible with its supersession date, and tell the dealers holding stock which variant is on their shelf. Channel mechanics matter too. eBay's seller guidance describes two paths — list against its catalog for prefilled compatible vehicles and automatic updates, or, when a part is not in the catalog, add compatible vehicles yourself by year, make and model with trim and engine optional. For a pre-1981 part the second path is the normal one, so the compatibility data buyers see is yours, and no better than your own records.

Often, and we would rather say so here than after you have paid. Application research is permanently your cost: it does not get cheaper as the order gets bigger, it cannot be bundled into a piece price, and no factory — at 200 pieces or 5,000 — will find the running change that moved your mounting hole. And if what you actually need is ACES and PIES data formatted for a marketplace or distributor feed, a US fitment-data service is the right vendor and we do nothing for you there. Taiwan Supplier Hub's structured catalog is automotive hard parts; it is not and never will be a pre-1981 application database.

What we are useful for is the other half — a tooled or fabricated part you have already decided to have made. Describe it in plain English on our request page: vehicle, years, body style, position, material, finish, and what donor sample you hold. Taiwan Supplier Hub returns a shortlist of at least three verified manufacturers that have confirmed they want the project within 14 days, US$99, refunded if we cannot deliver. If the conversation shows you have a data problem rather than a sourcing problem, we will say so before you pay rather than after.