Methodology: Sources, Decoding Rules and Price Sampling

1. Seiko technical guides and parts catalogs, for movement specifications, parts and case construction.
2. Period catalogs, for when a reference was sold, in which markets, and with which dials.
3. Seiko’s own published histories, for dates and firsts.
4. Documented collector and watchmaker research, such as forum threads with photos and serials, and long-running reference sites. We use these to cross-check, never as the only source for a number.
5. Our own photos and measurements. Whenever possible, case dimensions are measured on watches we’ve handled.

Every source is recorded with its title, link and the date we accessed it. If two sources disagree, the page shows both and says which one we follow and why.

If we can’t confirm a value, we leave it blank and say why. We don’t fill gaps with a plausible guess. Values waiting for a second source are labelled as unverified.

A Seiko case back usually carries three codes: the caliber and case number (for example 6309-7040), and a serial number.

• First digit of the serial: the last digit of the production year. A 7 can mean 1967, 1977 or 1987.
• Second character: the month. 1 to 9 for January to September, then 0 (sometimes written O) for October, N for November and D for December.
• The rest of the serial is not a production count. We don’t use it for dating.
• Seven-digit serials: most vintage serials have six digits, and that is where these rules are documented. Applying them to seven-digit serials is a working assumption that hasn’t been checked for every reference, so those results carry one less level of confidence.
• Narrowing the decade: each caliber was produced during a known window, so we drop the decades that fall outside it. A 6139 was made from 1969 to 1979, so a serial starting with 7 points to 1977.

Each result carries a confidence level:

• High: only one decade fits the caliber’s production window.
• Medium: two decades fit, but catalog dates or dial markings (such as “Water Proof” versus “Water Resist”) can narrow it down.
• Low: three or more decades fit, or the caliber’s production window is itself unverified.

Leaving out the caliber or case code lowers the confidence one level. Every result is a possible production date, not a guaranteed date or an authentication. The production date is not the purchase date either: some watches sat in shops for years before they were sold.

The authenticity check compares the codes you enter against each other and against what Seiko actually made:

• Does the serial’s year fall inside the caliber’s production window?
• Does the caliber on the case back match the case code’s family and the decoded year?
• Has this dial printing ever appeared on this reference?
• Does the combination match a known fake or “franken” pattern? We list a pattern only when at least two independent sources document it.

The result is a consistency check, not an authentication. Matching codes don’t prove a watch is original, because case backs and dials can be swapped. For a definitive answer, have a watchmaker who knows vintage Seiko open the case, or use a paid authentication service.

Price ranges are the part of this site most likely to be copied and least likely to be checked, so here is the full protocol.

1. Source: eBay sold listings, reviewed by hand, for one reference at a time, over the last 90 days. Chrono24 asking prices are recorded separately as an upper reference. Asking prices are never mixed into sold prices.
2. Sample size: 15 to 30 sales per reference. If we can’t find 15 in 90 days, we extend the window to 180 days and mark the range as a small sample.
3. Condition grades:
   • A, collector grade: original, serviced, with box, papers or the original bracelet.
   • B: serviced and original, without the box, papers or original bracelet.
   • C: running but unserviced, or with minor flaws.
   • D: project watches, or suspected frankens.
4. Excluded: parts and movements sold alone, modified watches, obvious frankens, listings too vague to grade, the wrong reference, and modern reissues or homages that use the vintage reference number in their title.
5. Calculation: low, median and high for each grade, including shipping, rounded to the nearest $10.
6. What you see: the date, the sample size and the window, printed next to every range. For example: “Sampled 2026-11-05, n=24, eBay sold listings viewed manually.”
7. Updates: monthly. If the median moves more than 15% from the previous sample, the page shows the direction.
8. What we keep: only the aggregate figures and our own grading. We don’t store listing titles, photos, seller names or links, and we never scrape eBay.

A single dealer’s asking price, or one auction result, is never presented as “the market.”

Photos are our own or used with permission and credited. We never use photos from eBay listings or dealer sites.

No corrections yet. Each one will be listed here with the date, the page, what changed and why.

We use AI tools for research and for drafting. Every fact on the site is checked against the source it cites, and anything we can’t verify is marked as pending until we can.