A DOL case number looks like G-100-26125-868956. Every part of it means something, and one part is commonly explained wrong.
The shape
G - 100 - 26125 - 868956
│ │ │ └── serial: one counter shared by every program, six digits, wraps at 999999
│ │ └── filing day: calendar year 2026, day 125 of the year (May 5)
│ └── a three-digit segment, see below
└── the letter: which program this filing belongs to
The letter is the program
This is the part that matters most, and it is measurable across every case in DOL's system:
| Prefix | Form | What it is |
|---|---|---|
G- | ETA-9089 | A PERM labor certification |
A- | ETA-9089 | A PERM filed on the older form |
P- | ETA-9141 | A prevailing wage determination request |
I- | ETA-9035 | An H-1B labor condition application (LCA) |
Checked against the full corpus, this is clean: every PERM case carries G- or A-, every wage request P-, and every LCA I-. There are no exceptions in 374,000 PERM cases, 68,000 wage requests and 220,000 LCAs.
G or A is the form generation, and the switch has a date
Older PERM cases start with A-, newer ones with G-. The changeover is sharp and measurable:
| Prefix | Filing dates in the data | Cases |
|---|---|---|
A- | 15 May 2017 to 31 May 2023 | 92,248 |
G- | 1 June 2023 onward | 281,691 |
That boundary is the day DOL's revised ETA-9089 took effect. So if your number starts with A-, your case was filed under the old form. Both are valid and both are looked up the same way.
The three digits in the middle
You will read in several places that this segment identifies the processing centre, Atlanta or Chicago. That may well be right, and it is not something this data can confirm or deny.
What can be said is what it is not. All four values appear across the same date range, so it does not encode time:
| Segment | Filing dates | Cases |
|---|---|---|
G-100 | Jun 2023 to Jun 2026 | 192,667 |
G-200 | Jun 2023 to Jun 2026 | 84,145 |
G-300 | Jun 2023 to Jun 2026 | 4,780 |
G-400 | Jun 2023 to Jun 2026 | 99 |
DOL doesn't publish a key for it, and nothing in the public record maps a value to a centre. It has no bearing on how your case is processed as far as anyone can demonstrate, so it isn't worth reading anything into.
The date segment, which is genuinely useful
The five digits after the middle segment are a date: the first two are the calendar year, the next three the day of that year. So 26125 is day 125 of 2026, which is May 5 (not the fiscal year, which is how this is usually explained; a fiscal reading would put it in February).
The serial after it is not per day and not per program. DOL runs one counter across PERM, prevailing wage requests and LCAs alike, so a PERM filed at 2 PM and a wage request filed at 2:01 PM get neighbouring numbers. It is six digits, zero-padded (070550, never 70550), and it wraps: on June 10, 2026 it ran up to 999997 and restarted at 000001. The counter advances roughly 2,000 to 3,000 a day, of which about a tenth are PERM.
This is why a filing date can be read straight off the number without looking anything up. It's right about 95% of the time and off by a day for the rest, because the counter rolls at a moment that doesn't perfectly align with the calendar.
Where to find your number
It's on the receipt DOL sends whoever filed, so your employer or its attorney has it. Asking is the quickest route.
If that isn't working, every one of these is searchable by employer, job title and filing month:
- PERM cases for
G-andA-numbers - Wage requests for
P-numbers - H-1B LCAs for
I-numbers
There's a fuller walkthrough in how to find a PERM case number by employer name.
Checking a number once you have it
Case status takes all three prefixes and works out the program from the letter. If the number isn't already held, it asks DOL directly at that moment, so a recent filing still resolves.
What this can't tell you
- What the middle segment means. DOL publishes no key, and this page won't guess.
- Whether a number is real before you check it. A wrong digit still makes a well-formed number, which is why a typo looks exactly like a missing case.
- Anything about cases before 2017. They're not in the published data.



