Skip to content
Trademark Valet

Research library · Guide

How similar do two marks have to be?

Lawyers call this: du Pont factor 1 · similarity of the marks

Short answer

Similar enough that a buyer who half-remembers one mark would think the other comes from the same company. The comparison is of whole marks — how they look, sound, and what they mean — not a letter-by-letter count.

There is no fixed threshold. The closer the goods or services, the less similarity it takes.

Why it matters

This is one of the two factors the USPTO weighs in every likelihood-of-confusion case. The other is how related the goods or services are. A strong showing that the marks differ can decide a case on its own, even when the goods are identical.

It is also the factor you control, because you choose the name and the design.

The governing law

The first du Pont factor asks about “the similarity or dissimilarity of the marks in their entireties as to appearance, sound, connotation and commercial impression.” In re E.I. du Pont de Nemours & Co., 476 F.2d 1357, 1361 (C.C.P.A. 1973); U.S. Pat. & Trademark Off., Trademark Manual of Examining Procedure § 1207.01(b) (May 2026) [hereinafter TMEP].

What the factor asks

The test is not whether someone could tell the marks apart side by side. It is whether the marks are close enough in overall impression that people who meet them would likely assume a connection between the two businesses. In re i.am.symbolic, llc, 866 F.3d 1315, 1323 (Fed. Cir. 2017) (quoting Coach Servs., Inc. v. Triumph Learning LLC, 668 F.3d 1356, 1368 (Fed. Cir. 2012)). The focus is the average buyer, who keeps a general impression of a mark, not a precise one. TMEP § 1207.01(b).

The marks are compared as they appear in the application and the registration, not as they appear on labels. A house mark you plan to add does not count. In re Shell Oil Co., 992 F.2d 1204, 1207 n.4 (Fed. Cir. 1993). A standard-character registration is not limited to any particular lettering or style. Cunningham v. Laser Golf Corp., 222 F.3d 943, 950 (Fed. Cir. 2000).

How the Board weighs it

The marks are compared in their entireties. But the Board may, for stated reasons, give more weight to the dominant part of a mark. In re Nat’l Data Corp., 753 F.2d 1056, 1058 (Fed. Cir. 1985). Several rules follow from that:

  • Added words rarely help. Adding a house mark or a descriptive word to someone else’s mark usually does not avoid confusion when the dominant part is the same. TMEP § 1207.01(b)(iii); In re Detroit Athletic Co., 903 F.3d 1297, 1304 (Fed. Cir. 2018).
  • Words usually outweigh designs. Buyers use the words to ask for the product. TMEP § 1207.01(c)(ii); In re Viterra Inc., 671 F.3d 1358, 1366 (Fed. Cir. 2012).
  • There is no correct pronunciation. An applicant cannot rely on how it intends the mark to be said. TMEP § 1207.01(b)(iv); Viterra, 671 F.3d at 1367.
  • Meaning counts. A foreign word can be compared with its English equivalent, and reversing the order of the same words may not change the impression. TMEP §§ 1207.01(b)(vi), 1207.01(b)(vii).
  • A weak shared element counts for less. A descriptive or weak term gets a narrower scope of protection, though even a weak mark is protected against a similar mark on closely related goods. TMEP § 1207.01(b)(ix); King Candy Co. v. Eunice King’s Kitchen, Inc., 496 F.2d 1400, 1401 (C.C.P.A. 1974).

The factor also slides against the second one. Where the goods are identical or virtually identical, less similarity between the marks is needed to support a finding of confusion. TMEP § 1207.01(b) (citing Viterra, 671 F.3d at 1363).

What evidence moves it

Mostly the marks themselves. Beyond them: dictionary entries and translations, and evidence that a shared term is common or descriptive in the field. The rules are the same in an ex parte appeal and in an opposition.

Common mistakes

  • Comparing the marks side by side, or counting shared letters.
  • Arguing from the logo you use when the application or the registration is in standard characters.
  • Assuming a disclaimed or descriptive word sets the marks apart.
  • Assuming a design will carry the mark. It can, but usually the words dominate.

What this library shows about similarity of the marks

38 of the 38 records in this library that code du Pont factors carry a finding on factor 1 (library as of 2026-09-19; a curated collection, not a random sample). It is the only factor present in every one.

The coding records it as favoring a finding of confusion in 21 and as favoring the applicant in 4. The other 13 are coded as significant or dispositive without a recorded direction. In each of the 4 coded for the applicant, the Board reversed the refusal.

A design outweighed shared words. The REDNECK RACEGIRL logo was allowed over RACEGIRL for overlapping clothing. The Board found that the design’s giant checkerboard double-R dominated the mark and made the wording hard even to notice, so the marks conveyed different overall impressions despite the shared RACEGIRL.

Meaning outweighed appearance. MARAZUL was refused for fish and seafood over the registered BLUE SEA. “Mar azul” is Spanish for “blue sea,” and the Board found that shared meaning outweighed the differences in how the marks look and sound.

Reversed words, same meaning. In an opposition, the Board refused Pharmavite’s NATURE MADE for foods over MADE IN NATURE, finding NATURE MADE a transposition of the same words conveying the same meaning, with goods identical in part.

Method. n = 38 records in which at least one du Pont factor was coded, drawn from a library of 178 Board decisions coded to a fixed schema as of 2026-09-19. 38 of the 38 carry a finding on factor 1. A record is counted when its coding carries a finding on the factor, whether that finding was neutral, weighed for or against confusion, or drove the result. A record is one application or proceeding number, so one opinion that decides three consolidated oppositions counts as three records. This is a curated library, not a random sample of all TTAB decisions. The counts describe this library. They are not a probability and not an estimate of any wider population.

Important decisions

See all 178 decisions →

All 38 decisions coded on this factor →

What to do next

If you are choosing a name, run the comparison the way the Board does. Say both marks aloud. Translate any foreign word. Strip out descriptive words and house marks and compare what is left. Then compare the result against the closest registrations for the same or related goods. You can do this first pass yourself.

If you already have a refusal, read how the examining attorney identified the dominant part of each mark. That choice drives the rest of the analysis.

A close mark on the same goods, or a refusal in hand, is where an attorney’s read earns the fee, because the argument turns on weighting judgments and on the record you build. If you have that refusal, Valet Law, PLLC can review the examining attorney’s comparison and tell you whether the marks argument is worth making.

Attorney review is provided by Valet Law, PLLC, a law firm separate from Trademark Valet, LLC. Keep confidential facts out of public tools. Your first message to Valet Law should identify the parties and the general issue only. Wait until Valet Law confirms it can discuss the matter before sending confidential details.

Update log

2026-09-19Page created. Corpus counts computed from 178 analyzed decisions.

Who checked this

Reviewed and approved by Brandon Leavitt on 2026-10-02.

General information about how trademark law works. It is not legal advice about your situation, and reading it does not make anyone your lawyer.