Our search's scope and limits

What it does check

• The federal USPTO trademark register — active and pending applications and registrations.
• The Nice class(es) you select, plus classes USPTO's own data shows are commonly related to them.
• Textual similarity, and phonetic similarity using the same category of algorithm USPTO's own examiners use to flag "sounds-alike" marks.
• For a logo search specifically: USPTO's own Design Search Code system, matched against your uploaded image.

What it doesn't check

• Common-law rights — marks in real use that were never federally registered. These can still block your application or give someone else grounds to oppose it.
• State trademark registrations — each state runs its own, separate register; this search doesn't touch them.
• A full §2(d) likelihood-of-confusion analysis — the multi-factor legal test (the DuPont factors) an examining attorney or a court actually applies. This search surfaces mechanical similarity signals for you to weigh; it doesn't weigh them for you.
• Design-mark similarity in a word-mark search, and vice versa — search by whichever type actually matches what you're protecting, and consider running both if your mark has real elements of each.

What that means for you

A clean result narrows real risk — it doesn't eliminate it, and it isn't a promise your application will register. Treat this the way you'd treat any preliminary search: a genuinely useful first step, not the final word.