Section 2(d) refusals: what “likelihood of confusion” actually means
A Section 2(d) refusal does not say your mark is identical to someone else's. It says an examining attorney at the USPTO believes consumers could be confused about who is behind your product or service. That is a lower bar than "the same," and it is why founders and artists are so often blindsided by one.
Here is the statute. Section 2(d) of the Trademark Act, codified at 15 U.S.C. § 1052(d), bars registration of a mark that "so resembles a mark registered in the Patent and Trademark Office, or a mark or trade name previously used in the United States by another and not abandoned, as to be likely, when used on or in connection with the goods of the applicant, to cause confusion, or to cause mistake, or to deceive."
Three words in that sentence do the work: resembles, likely, and confusion. Not "copies." Not "certain." Not "identical."
The DuPont factors are a checklist, not a scorecard
Examining attorneys analyze likelihood of confusion using the factors from In re E. I. du Pont de Nemours & Co. The USPTO's Trademark Manual of Examining Procedure (TMEP) — the internal rulebook examiners follow — collects this analysis at TMEP § 1207.01.
Two factors are always in play:
The similarity or dissimilarity of the marks in their entireties as to appearance, sound, connotation, and commercial impression.
The relatedness of the goods or services as described in the application and the cited registration.
The rest — trade channels, purchasing conditions, the number of similar marks already in the field, the fame of the cited mark, consent agreements — come in only when there is evidence about them. The TMEP is explicit that "not all of the DuPont factors are relevant to every case" and that "any single factor may control a particular case." There is, in its words, "no litmus rule which can provide a ready guide to all cases."
So this is not arithmetic. You do not win five of nine and take the round.
Factor one: the marks, compared the way people actually remember them
Under TMEP § 1207.01(b), marks are compared in their entireties. You cannot slice off the part you dislike and argue the remainder. But examiners may reasonably give "more or less weight to a particular feature," as long as the conclusion rests on the whole mark.
Two consequences matter for how you argue.
Recollection, not side-by-side. The test looks to "the recollection of the average purchaser who normally retains a general, rather than specific, impression of trademarks." Consumers do not hold two labels up to the light. They half-remember one and encounter the other three weeks later.
Sight, sound, or meaning — any one can be enough. Similarity in appearance, in phonetic sound, or in connotation can each support a refusal. TMEP § 1207.01(b) covers similarity in appearance at (b)(ii), phonetic equivalents at (b)(iv), and similarity in meaning at (b)(v). It also addresses the doctrine of foreign equivalents at (b)(vi) — a mark in another language may be compared against its English translation.
This is the answer to the most common client question: how can they refuse me when the marks look completely different? Because "different" to the eye is not the test. Two marks spelled nothing alike can sound alike. Two marks that sound nothing alike can mean the same thing.
Factor two: the goods and services as written, not as sold
This is where self-filers lose ground before the argument starts.
Under TMEP § 1207.01(a)(i), "the goods or services do not have to be identical or even competitive to find a likelihood of confusion." They "need only be related in some manner and/or the circumstances surrounding their marketing are such that they could give rise to the mistaken belief that they emanate from the same source."
And under TMEP § 1207.01(a)(iii), the scope of each party's goods is determined "on the basis of the goods or services recited in the application and cited registration" — not what either business actually sells today. If your identification says "clothing," you are compared as a clothing company, even if you only make one hoodie.
There is a sliding scale built in: "The more similar the marks at issue, the less similar the goods or services need to be to support a finding of likelihood of confusion." Very close marks can be refused across a surprisingly wide gap in product categories. TMEP § 1207.01(a)(iv) also confirms there is no "per se" rule — no category pairing is automatically related or automatically not.
The thumb on the scale
One rule you should know before you decide how to respond. TMEP § 1207.01(d)(i) states: "If there is any doubt as to whether there is a likelihood of confusion, that doubt must be resolved in favor of the prior registrant."
Close calls do not default to you. Plan accordingly.
Related points from the same section: the absence of actual confusion is addressed at § 1207.01(d)(ii), and you generally cannot attack the validity of the cited registration inside your own application — TMEP § 1207.01(d)(iv) treats collateral attack on a registration as improper in an ex parte proceeding. That fight, if it is worth having, belongs in a separate proceeding before the Trademark Trial and Appeal Board.
How Section 2(d) refusals get answered
Four broad paths. They are frequently combined.
1. Argue the factors. Attack similarity, relatedness, or both, with evidence. One of the more useful angles is at TMEP § 1207.01(d)(iii): "A large number of active third-party registrations including the same or similar term . . . may be given some weight to show that a mark . . . has a normally understood descriptive or suggestive connotation, leading to the conclusion that the term . . . is relatively weak." If a term is crowded, each user gets a narrower zone. Sophistication of purchasers (§ 1207.01(d)(vii)) and fame of the cited mark (§ 1207.01(d)(ix)) are also live considerations where evidence supports them.
2. Amend the identification. Because the comparison runs on the recited goods and services rather than real-world sales, narrowing or restricting your identification changes the record the examiner is comparing. This is a real tool, and it has a real cost: you are permanently shrinking what your registration covers.
3. Consent agreements. The registrant can consent to your registration, and that consent goes into the record. TMEP § 1207.01(d)(viii) is the governing section. Weight depends entirely on substance. A "naked" consent — bare permission plus a conclusory statement that confusion is unlikely — carries little. What the TMEP looks for is a detailed agreement: the parties' stated reasons for believing confusion is unlikely, differences in trade channels or fields of use, specific arrangements the parties will undertake to avoid confusion, an agreement to cooperate if confusion arises, and any history of concurrent use.
4. Coexistence terms. In practice, the consent filed at the USPTO is often carved out of a broader private coexistence contract between the two businesses covering territory, product lines, presentation, and enforcement. The USPTO weighs what is in the record. The contract governs the parties.
The clock
Do not lose the application to a calendar. For most pre-registration office actions, the USPTO requires a response within three months of the issue date, with an option to request one three-month extension for a fee. Applications filed under Section 66(a) through the Madrid Protocol get six months and cannot extend. Confirm the period stated in your own office action — the USPTO says so directly.
A 2(d) refusal is an opening position by an examining attorney working from a written record. It is answered with evidence and precision, not indignation.
This article is general information about USPTO procedure, not legal advice, and does not create an attorney-client relationship. Every application turns on its own facts.
Sources:15 U.S.C. § 1052 (Cornell LII) · TMEP § 1207.01 — Likelihood of Confusion · TMEP § 1207.01(a) — Relatedness of the Goods or Services · TMEP § 1207.01(b) — Similarity of the Marks · TMEP § 1207.01(d) — Miscellaneous Considerations, incl. (d)(i), (d)(iii), (d)(viii) · USPTO — Response time period