Bates stamping and pagination rules every immigration paralegal should know

The most common mistake is restarting numbering inside each exhibit, producing three separate page 1s in one binder. When the cover letter cites page 12, the adjudicator has to work out which exhibit's page 12 is meant.

Number continuously across the entire assembled binder from the first page of the cover letter to the last page of the last exhibit. Exhibit tabs and the index provide the structure; Bates numbers provide the addresses.

A good prefix identifies the matter without depending on anything that might change. Beneficiary surname plus form type plus sequence works well: DOE-NIW-0001. Firm-code schemes work if the firm code is stable. Date-based prefixes age badly and break when a matter is refiled.

Use four digits minimum. Merit-based binders routinely exceed three hundred pages and EB-1A filings often pass six hundred; a three-digit scheme forces a mid-matter change, which is worse than starting wide.

• DOE-NIW-0001 — beneficiary, category, sequence. Stable and readable.
• SMITH-I140-0001 — beneficiary, form, sequence. Good for multi-petition beneficiaries.
• Avoid 2026-03-0001 — date prefixes break on refiling.
• Avoid restarting per exhibit — citations become ambiguous.

Bottom right is the convention, inset far enough from the edge to survive printing and scanning. The practical constraint in immigration work is that many source documents already carry stamps, apostilles, notary seals or agency markings in the corners. A stamp that lands on top of a notarial seal creates an exhibit that looks altered.

Check the corners of scanned foreign documents specifically. Where the standard position is occupied, move the stamp for the whole binder rather than for one page — consistency matters more than the specific corner.

An exhibit gets added two days before filing. The temptation is to give it a letter suffix — 0142A, 0142B — and leave everything else alone. This works until the cover letter cites a range that now spans the insertion, at which point the range is wrong and nobody notices.

Regenerate the entire numbering, then regenerate the index and every citation against the new range. Done by hand this is an hour of careful work under deadline. Done mechanically it is instant, which is the whole argument for taking it out of human hands.