On June 29, 2026, the Supreme Court granted certiorari (opens in new tab) in RiseandShine Corp. v. PepsiCo, Inc., No. 24-1016. If you practice trademark law, this is the case you will be hearing about for the next year. For the first time, the Court has agreed to look inside the likelihood-of-confusion test itself. The Court’s modern trademark cases have mostly circled the test: protectability, trade dress, tacking, preclusion, remedies, registration, extraterritoriality, expressive-use filters. This one asks how the test is operated, factor by factor, and by whom. (Disclosure: after this post first ran, I filed an amicus brief in support of neither party.)
A quick orientation. RiseandShine brews canned nitro cold brew under the mark RISE. In 2021, PepsiCo launched Mtn Dew Rise Energy, and RiseandShine sued on a reverse-confusion theory: Pepsi’s scale would swamp the smaller, senior RISE mark, and shoppers would come to think RiseandShine’s coffee came from Pepsi. The district court heard the evidence, found that strength tilted slightly toward RISE, and entered a preliminary injunction. The Second Circuit vacated it. One of the two errors it found was legal: strength, in that circuit, carries “a considerable component of law,” and the panel graded the mark for itself, finding RISE inherently weak. That inherent-weakness grade became law of the case. It bound the district court on remand, PepsiCo won summary judgment, and the Second Circuit affirmed in 2024.
The question presented is a single line: whether trademark strength is a question of fact in a likelihood-of-confusion analysis under 15 U.S.C. ยง 1114. As the petition tells it, twelve circuits treat strength as a question of fact, typically for a jury. The Second Circuit stands alone in grading it as law.
Two things make the grant unusual. First, the Solicitor General, invited to weigh in, told the Court the Second Circuit was wrong and still recommended denial, reasoning that the error was unlikely to change the outcome, in this case or more generally, and that reviewing one sub-factor alone would not clarify the test. The Court granted anyway. Second, both sides are arguing about the jury’s province, and that is where the empirical record has something to say.
The Question Beneath the Question
Put the two positions side by side and notice what they share. The petitioner wants strength in the jury box. The Second Circuit wants it on the bench. Both assume a trial is coming. My new essay, The Missing Jury: Fact, Law, and Who Actually Decides Trademark Confusion, measures that assumption against the docket.
Across 10,977 federal likelihood-of-confusion opinions, 7,050 resolve the confusion question on the merits. Judges decided 97.96% of those. Juries decided 2.04%, which is 144 opinions. The strength factor specifically, the dial the Court has now agreed to allocate, reached a jury 69 times out of 3,383 gradings. And the opinion data are, if anything, generous to the jury: the Administrative Office’s own termination tables show that in fiscal 2025, twenty-seven trademark cases in the entire federal system reached a jury, out of 3,851 terminations. That is 0.70%.
Academics have been counting for twenty years. Barton Beebe’s 2006 study coded five procedural postures, none of them a jury verdict. A 2010 replication in the Southern District of New York coded the same way. Daryl Lim ran the count in 2022 and said it plainly: the conventional wisdom that confusion cases “generally require jury determination” does not survive the data. My corpus extends the count more than thirtyfold over the largest earlier study, and it measures the jury’s share directly.
What the Court Is Actually Allocating
If the jury decides one case in fifty, then “question of fact for the jury” is, in operation, a rule about summary judgment. It tells the judge that strength belongs to the hypothetical reasonable jury and may be taken from it only when no reasonable factfinder could disagree. “Question of law” tells the judge the grade is hers, portable across cases and reviewable de novo, which is exactly the mechanism at work in RiseandShine itself: an inherent-strength grade made on an injunction record, relabeled as law, and insulated from later factual argument.
Either way, the decision-maker being regulated is a judge. The real choice is between standards of judicial self-restraint. The merits briefs owe the Court two answers: what a reasonable-jury screen means for a factor courts grade intuitively, and what a “legal” strength grade is made of when no two records present the same mark.
There is also a question hiding inside the question. “Strength” bundles two different jobs: assessing the mark on a record (how distinctive is RISE, on this evidence) and deciding what that assessment does inside the multifactor analysis (how the factors combine). Those two jobs may not deserve the same label. A Court looking for a clean line could treat the assessment as fact and the composition of the test as law, and this case gives it a vehicle to say so. Whether the merits teams invite that move is worth watching.
The full essay is published at 2026 U. Ill. L. Rev. Online 64 (opens in new tab) and is also on SSRN (opens in new tab). It is a fast read as these things go. Merits briefing is under way: the petitioner filed on September 8, six amicus briefs (including the United States) followed on September 15, and the respondent’s brief is due November 4. I will check back in when argument is set.
Prof. Reichert