Bolar Safe Harbor: 6 Essential Rules Under 35 USC 271(e)(1)

A competitor’s patent does not expire for four more years, and your team needs to run the studies now so the product can launch the day it does. Touch the patented compound today and you are, on the plain text of the statute, an infringer. The Bolar safe harbor is what makes that work legal — broader than most engineers assume, narrower than most business teams hope. The gap between those beliefs is where infringement suits are born. Here is what 35 U.S.C. § 271(e)(1) protects, the decisions that set its edges, and six rules for staying inside it.

What the Bolar Safe Harbor Actually Protects

Start with the text, because almost every argument about this exemption is really an argument about eight words in it. Section 271(e)(1) says it is not an act of infringement to make, use, offer to sell, or sell within the United States, or import into the United States, a patented invention “solely for uses reasonably related to the development and submission of information under a Federal law which regulates the manufacture, use, or sale of drugs or veterinary biological products.” You can read the provision in full at Cornell’s Legal Information Institute.

Three features do the work. First, the exemption is an activity test, not a party test — nothing in it is limited to generic manufacturers, and nothing in it turns on whether you are a for-profit company. Second, it protects conduct that would otherwise be textbook infringement. Third, everything depends on reasonably related, a phrase Congress never defined and the courts have spent forty years mapping.

What the exemption is not: a license to sell. It shelters research and regulatory-submission activity. The moment the activity becomes ordinary commerce, the shelter is gone.

Roche v. Bolar: The Case That Forced Congress to Act

The doctrine carries a defendant’s name because that defendant lost. In Roche Products, Inc. v. Bolar Pharmaceutical Co., 733 F.2d 858 (Fed. Cir. 1984), a generic manufacturer ran bioequivalence testing on a patented drug before the patent expired, so it could launch at expiry. Bolar argued experimental use. The Federal Circuit refused, calling the common-law exception “truly narrow” and telling Bolar where to take its complaint: it is the role of Congress, the court said, to maximize public welfare through legislation.

Congress took the invitation within months. The Drug Price Competition and Patent Term Restoration Act of 1984 — Hatch-Waxman, Pub. L. 98-417 — added two provisions that still define the field:

• § 271(e)(1) — the safe harbor, which overruled the result in Roche and legalized pre-expiry regulatory research.
• § 271(e)(2) — the “artificial” act of infringement, which makes filing certain drug applications itself a suable event, giving the patent owner a forum before launch.

That pairing is the bargain: generics got to prepare early, innovators got an early cause of action and restored patent term. The term-restoration half is covered in our guide to patent term extension under 35 U.S.C. § 156.

How Far Reasonably Related Really Reaches

For twenty years the open question was whether the exemption covered only the narrow, ANDA-style testing that Roche involved. The Supreme Court answered in Merck KGaA v. Integra Lifesciences I, Ltd., 545 U.S. 193 (2005) — unanimously, in an opinion by Justice Scalia — and the answer was expansive. The statute, the Court held, provides a wide berth for the use of patented compounds in activities related to the federal regulatory process.

Three points from Merck matter in practice:

1. Preclinical work counts. The exemption reaches preclinical studies of safety, efficacy, and mechanism of action — not just human trials.
2. You do not have to be right. Protection does not evaporate because the compound fails, or because the data never reaches the FDA. What matters is whether there was a reasonable basis to believe the compound tested could be the subject of an FDA submission.
3. Aimless research is out. “Wide berth” is not “no berth.” Basic research untethered from any intent to develop a regulated product falls outside the provision.

One limit worth knowing because almost no one mentions it: Merck expressly declined to decide whether the safe harbor covers research tools — patented instruments used to find drugs rather than compounds destined for a submission. If your exposure runs through a research-tool patent, treat that question as open, not settled.

It Is Not Just Generic Drugs — and the Statute Never Says Devices

Read § 271(e)(1) again and you will notice something odd: it mentions drugs and veterinary biological products. It never mentions medical devices. Yet device makers rely on the exemption constantly. Why?

Because of Eli Lilly & Co. v. Medtronic, Inc., 496 U.S. 661 (1990). The Supreme Court held that the safe harbor reaches premarket activity undertaken to develop and submit information for device approval under the Federal Food, Drug, and Cosmetic Act. The district court had said devices were outside the statute; the Supreme Court held they are in.

This is worth flagging because much recent commentary points at the Federal Circuit’s 2024 Edwards decision as the device authority. It isn’t. Medtronic put devices inside § 271(e)(1) thirty-four years earlier; the newer cases only apply that rule. If you are briefing this, cite the 1990 decision — it is the load-bearing one.

Where the Safe Harbor Runs Out: Routine Commercial Testing

Here is the boundary that trips people, and the one most published explainers state incorrectly because they were written before the law settled.

In 2012, a Federal Circuit panel held that a generic maker’s post-approval quality-control testing was sheltered, reasoning that it generated records the FDA required for continued approval. Many articles still repeat that rule. It is no longer good law. On the fuller record, the Federal Circuit came out the other way in Momenta Pharmaceuticals, Inc. v. Teva Pharmaceuticals USA, Inc., 809 F.3d 610 (Fed. Cir. 2015) — the consolidated appeal often cited loosely as Momenta v. Amphastar. It held that routine quality-control testing of each batch, as part of post-approval commercial production, is not “reasonably related to the development and submission of information” to the FDA. The Supreme Court denied review.

The operative word is routine. The distinction is not simply pre-approval good, post-approval bad. It is closer to this: information developed to obtain or change a regulatory determination tends to be sheltered; testing performed over and over as a condition of selling product you already sell is ordinary manufacturing, and the exemption does not cover ordinary manufacturing.

The practical translation: ask what the data is for. If the honest answer is “because we ship it,” assume you are outside the safe harbor. That question belongs in the same file as your freedom-to-operate search, not in a separate regulatory silo.

The 2024 Edwards Decision and the Word Solely

The most recent significant application is Edwards Lifesciences Corp. v. Meril Life Sciences Pvt. Ltd. (Fed. Cir. No. 22-1877, decided March 25, 2024). Meril imported two transcatheter heart valve systems into the United States for a medical conference. The devices were never taken out of the bag. Months later, Meril submitted a premarket approval application to the FDA. Edwards sued over the importation.

The Federal Circuit affirmed summary judgment of non-infringement: the importation was reasonably related to submitting information to the FDA, and therefore sheltered. The panel split 2-1, with Judge Lourie dissenting, and the court denied rehearing en banc in August 2024.

The fight was over the statutory word solely. Does an importation lose protection if the importer also had a commercial motive? The majority’s approach is objective: it asks whether the conduct was reasonably related to a regulatory submission, not whether the actor also harbored business hopes. That is a defendant-friendly reading, and it is the rule in force. Two cautions. A dissent this pointed means the question is live. And the Supreme Court’s later refusal to hear the case is not an endorsement — a denial of certiorari decides nothing on the merits, whatever the client alerts imply.

The Common-Law Experimental Use Exception Will Not Save You

Ask a researcher why they can use a patented invention and you will often hear “it’s for research.” There is a judge-made experimental use exception in U.S. patent law. It is, for practical purposes, dead.

Madey v. Duke University, 307 F.3d 1351 (Fed. Cir. 2002), is the reason. A physicist sued his former university over lab equipment covered by his patents. Duke argued it was a non-profit doing science. The Federal Circuit held that non-profit status is not the question. So long as the act is in furtherance of the alleged infringer’s legitimate business — and a university’s business includes research, teaching, and chasing grants — the defense fails. It survives only where the use is “solely for amusement, to satisfy idle curiosity, or for strictly philosophical inquiry.”

Read that standard honestly and almost nothing an institution does qualifies. Which is why § 271(e)(1) matters so much: for regulated products, the statutory exemption is the only real shelter there is.

6 Rules for Staying Inside the Bolar Safe Harbor

What this looks like as working practice:

1. Tie every study to a submission. The test is objective, but objectivity is proved with documents. A contemporaneous record of which regulatory question a study answers is the cheapest evidence you will ever create.
2. Do not rely on failure being forgiven twice. Merck protects work on compounds that wash out — but only where there was a reasonable basis at the time to think it could support a submission. Reconstructing that basis in litigation is a bad plan.
3. Quarantine routine production testing. After the 2015 Momenta decision, batch testing done because you are selling product is exposed. Separate it, in process and paperwork, from development testing.
4. Do not import commercial volumes and call it research. Two demo units for a conference sat inside the exemption in Edwards. A pallet with a purchase order will not.
5. Treat research tools as an open question. The Supreme Court reserved it. If your risk runs through a tool patent rather than a compound, get a license or a real opinion.
6. Never substitute the common-law exception. If your product is not FDA-regulated, § 271(e)(1) does nothing for you, and Madey means research framing will not fill the gap.

One point that gets missed: the safe harbor is a defense to infringement, not a shield against a patent’s existence. It ends the day the patent does. If your launch plan depends on a patent being weaker than it looks, that is a job for a patent invalidation search. And once a product is sold, a different doctrine takes over: patent exhaustion.

How PerspireIP Can Help

Safe-harbor questions are rarely decided on doctrine alone — they are decided on what the prior art shows and what your records prove. PerspireIP builds the evidence base: freedom-to-operate analysis around a development program, invalidity and prior-art searching on the patents that stand between you and a launch date, and landscape work that tells you which patents actually expire when. If you are planning research against a live patent and want to know where your exposure really sits, talk to our team.

Frequently Asked Questions