A non-disclosure agreement (NDA) is usually the first contract a university signs in a technology-transfer relationship — before a license, before a sponsored research agreement, sometimes before either side knows whether there’s a deal to be had at all. Its job is narrow: let both sides talk candidly about unpublished, proprietary, or otherwise non-public information without either one losing control of it. Everything else — who owns what gets invented, who can publish what, who pays for what — is deliberately left to whatever agreement comes next.
That narrowness is exactly where research institutions run into friction that a standard commercial NDA template doesn’t anticipate. A corporate NDA assumes both signatories can simply stop talking about a topic indefinitely if the deal falls through. A university signatory usually can’t: faculty have an academic-freedom interest in publishing their own results, a funder may have its own disclosure timeline, and the confidential information often sits next to inventions the institution may itself have rights in under the Bayh-Dole Act. This guide covers how NDAs are actually used in a research and technology-transfer context, what belongs in one, what should be explicitly excluded from one, and where an NDA stops and a different agreement — a sponsored research agreement (SRA) or a material transfer agreement (MTA) — needs to take over.
When a research institution uses an NDA
Outside of research, NDAs mostly show up around employment and vendor relationships. Inside a university or research institute, a technology transfer office (TTO), sponsored programs office, or industry-engagement office signs or reviews NDAs for several distinct, recurring scenarios:
- Evaluating a potential license. Before a company can meaningfully assess whether to license a university invention, it typically needs more detail than the public patent application or a one-page technology summary provides — unpublished data, additional experimental results, or a prototype’s specifications. An NDA lets the TTO share that detail during due diligence without prejudicing the institution’s ability to license the technology to someone else if the deal doesn’t close.
- Pre-agreement partnership discussions. Before a company and a university negotiate a sponsored research agreement, a master research agreement, or a consortium membership, both sides usually need to exchange enough information to scope the collaboration — what each party’s existing capabilities and constraints are, what problem the company wants solved, what background technology the university would be building on. An NDA covers that scoping conversation; it is not itself the research agreement, and it doesn’t authorize any research to actually start.
- Visiting researcher and collaborator confidentiality. When a visiting scholar, sabbatical visitor, or short-term collaborator will be exposed to a lab’s unpublished data, a sponsor’s proprietary materials, or another party’s confidential information as a condition of the visit, institutions frequently require a confidentiality undertaking alongside the visiting scholar agreement that governs the visit itself.
- Vendor, contractor, and core-facility relationships. A contract manufacturer, an outside testing lab, or a specialized service provider brought in to support a research project may need to see proprietary materials or methods to do the work, without becoming a research collaborator or acquiring any claim to resulting IP.
In every one of these cases, the NDA’s function is the same: protect a defined category of information for a defined purpose and a defined period, without creating any of the substantive rights — IP ownership, funding obligations, publication terms — that belong in whatever agreement follows.
Mutual vs. one-way NDAs in a research context
NDAs come in two structures, and which one applies depends on who’s actually disclosing information:
One-way (unilateral) NDA — only one party discloses confidential information; the other only receives it. This is common when a company shares proprietary background technology, formulation data, or business plans with a university team purely so the team can evaluate a potential collaboration, without the university sharing anything comparably sensitive in return. It’s also the shape used when a university licenses out a technology and shares supplementary unpublished data with a prospective licensee during diligence — the flow of confidential information is essentially one-directional.
Mutual (two-way) NDA — both parties disclose and both parties receive confidential information, and both are bound by the same obligations with respect to what they receive. This is the more common structure once a real collaboration is being scoped: a company describes its proprietary process or product roadmap while the university describes unpublished preliminary results or in-progress methods, each testing whether the other’s information changes what a joint project should look like. Two universities discussing a multi-institution collaboration, or an institution and a foreign partner scoping a joint proposal, also typically use a mutual NDA, since both sides are bringing non-public information to the table.
A practical point research administrators run into: a company will sometimes propose a one-way NDA (protecting only its own information) even in a situation where the university will also be sharing something non-public — for example, unpublished preliminary data used to justify why the university’s approach is worth partnering on. If the university is disclosing anything it wants protected, a one-way NDA drafted from the company’s side won’t cover it, and the institution needs to either negotiate mutual terms or decline to share beyond what’s already public until the agreement is corrected.
What a research-context NDA typically covers
A well-drafted research NDA, whatever its purpose, generally defines:
- What counts as confidential information — usually information marked or otherwise identified as confidential at the time of disclosure, sometimes with a short window (commonly 30 days) to confirm anything shared orally in writing.
- The permitted purpose — the single reason the information may be used (e.g., “solely to evaluate a potential license” or “solely to scope a proposed sponsored research collaboration”), which is the clause that keeps an NDA from being read as authorizing anything beyond that evaluation.
- Who may see it — typically limited to employees, faculty, and sometimes named students or collaborators who have a need to know and are themselves bound by confidentiality obligations, often with a requirement that the university not further share the information outside the evaluating team without consent.
- The term of confidentiality — how long the receiving party must keep the information confidential after receipt, independent of how long the agreement itself runs; in institutional practice this commonly runs three to seven years from disclosure (see the CASRAI Dictionary’s entry on the confidentiality agreement (NDA) for the underlying definition).
- Return or destruction obligations — what happens to confidential materials, samples, or documents if the discussions end without a further agreement.
What it typically excludes — and why the exclusions matter more in research
Every properly drafted NDA carries a standard set of exclusions from the confidentiality obligation: information that was already known to the receiving party before disclosure, information that becomes public through no fault of the receiving party, information independently developed without reference to the disclosure, and information rightfully received from a third party without a duty of confidentiality. These exclusions exist in corporate NDAs too. In a research institution, two of them carry extra weight.
Pre-existing and background IP. A university’s NDA needs to be explicit that signing it does not transfer, license, or encumber any intellectual property either party already owns or independently develops — including inventions the institution may go on to elect title to under the Bayh-Dole Act (35 U.S.C. § 200 et seq.) for federally funded work. An NDA is not a license and should never be treated as one; if a discussion under an NDA leads to an actual license, that license is a separate, subsequently negotiated agreement, and the NDA itself should say so plainly rather than leaving the point to be inferred.
Publication rights — the genuine research-specific wrinkle. This is the clause that most distinguishes a research NDA from a standard business NDA, and it’s the one general-purpose legal templates rarely get right for an academic signatory. A corporate NDA typically assumes indefinite silence about the covered subject matter is fine for both sides. A university researcher usually cannot accept that: publication of research results is central to an academic career, to funder compliance (many federal awards carry their own publication and data-sharing expectations), and to the institution’s own mission. A properly scoped research NDA should:
- Restrict its confidentiality obligation to the specific information actually disclosed under the agreement — not to the researcher’s own independently generated results, methods, or ideas, even where those overlap topically with the disclosed information.
- Make clear that a bare evaluation-stage NDA does not itself impose any publication delay or pre-publication review obligation. Publication-delay and sponsor-review clauses (commonly a 30–60 day review window, sometimes extended for patent filing) belong in the sponsored research agreement that governs actual funded work — see CASRAI’s guide to industry-university research partnerships — not in the preliminary NDA that precedes it. Conflating the two can leave a researcher unintentionally bound to indefinite silence over work that was never actually sponsored.
- Preserve the institution’s ability to file a patent application covering its own inventions arising independently of the disclosure, since an NDA restricting disclosure of a third party’s information should not be read as restricting the university’s own patenting activity on its own IP.
Where an NDA is a preliminary step to actual sponsored research, the practical answer is not to solve publication rights inside the NDA at all — it’s to keep the NDA narrowly scoped to the diligence conversation and negotiate the substantive publication, IP, and delay terms in the SRA that follows, where they belong and where they can be negotiated with the full picture of what’s being funded.
NDA vs. sponsored research agreement vs. material transfer agreement
These three agreement types are the ones research administrators most often see confused with one another, because all three can involve “sharing something proprietary with an outside party.” What they actually govern is different enough that using the wrong one — or trying to make one do the job of another — creates real gaps.
| Agreement | What it governs | Does research get performed under it? | Typical IP/publication terms |
|---|---|---|---|
| NDA | Confidential exchange of existing information (data, materials descriptions, business terms) for a defined evaluation purpose | No — evaluation and discussion only | None; IP ownership and publication are addressed by whatever agreement (if any) follows, not by the NDA itself |
| Sponsored research agreement (SRA) | A defined, funded research project — scope of work, deliverables, budget, reporting | Yes — this is the agreement that authorizes and funds the actual research | Negotiated directly: background vs. foreground IP ownership, license options, and a publication-delay/sponsor-review clause (see industry-university research partnerships) |
| Material transfer agreement (MTA) | Transfer of tangible research materials (cell lines, reagents, compounds, model organisms) between institutions, with use and redistribution restrictions | Sometimes — the recipient typically does research using the material, but under restrictions set by the MTA rather than a funded scope of work | Use restrictions on the material itself, publication notice requirements, and often IP terms covering inventions made using the material (see the MTA process guide) |
A useful way to keep the three straight: an NDA protects information that already exists, an MTA governs a tangible thing that moves between labs, and an SRA authorizes new work to happen with funding attached. It’s common for a relationship to pass through more than one of these in sequence — an NDA to scope the conversation, then an MTA to move a reagent needed for a feasibility test, then an SRA once both sides commit to funded work — and each should be scoped to only what it actually governs rather than stretched to cover the next stage.
Negotiation points research administrators should watch for
- Signature authority. Most institutions restrict who may bind the university to a confidentiality obligation — typically the TTO, sponsored programs office, or general counsel, not an individual faculty member. A faculty member signing a vendor’s or sponsor’s boilerplate NDA directly can commit the institution to terms — an overly long confidentiality term, an unfavorable definition of confidential information, or a publication restriction — that were never reviewed against institutional policy. Routing every incoming NDA through the office with signature authority, even ones that look routine, is standard TTO guidance.
- Term length. A shorter confidentiality term (often two to five years from disclosure) is generally appropriate for ordinary technical/business information; trade-secret-level information may reasonably justify a longer term, but an indefinite or unusually long term should prompt scrutiny of what’s actually being protected.
- Residual knowledge clauses. Some NDAs (more common when proposed by a corporate counterparty) include a clause allowing the receiving party to use information retained in the unaided memory of individuals who were exposed to it. These clauses are contested precisely because they can functionally erode the confidentiality obligation for anyone with a good memory; institutions should understand what’s being asked before accepting one.
- Export control overlap. If the information or the counterparty involves controlled technology, technical data, or non-U.S. persons, an NDA alone doesn’t resolve export-control exposure — that’s a separate screening question. See CASRAI’s guide to export control (EAR/ITAR) and international research collaboration for how that review interacts with confidentiality obligations involving foreign nationals or foreign partners.
- No warranty of accuracy. Research NDAs typically disclaim any warranty that the disclosed information is accurate, complete, or non-infringing — the exchange is for evaluation purposes, and neither side should be read as vouching for the disclosed material’s correctness.
Frequently asked questions
Does a research NDA affect a researcher’s right to publish?
A properly scoped NDA shouldn’t. It should restrict disclosure of the specific confidential information received, not the researcher’s own independently generated results or ideas. Publication-delay obligations (a sponsor review period before submitting a manuscript) belong in a sponsored research agreement, not in a preliminary evaluation-stage NDA — if a researcher is asked to sign an NDA that includes an indefinite publication restriction, that’s a sign the document has drifted beyond what an NDA should cover and needs review by the TTO or sponsored programs office before signing.
Can a faculty member sign an NDA directly, without routing it through the technology transfer office?
Institutional policy varies, but most universities restrict signature authority for confidentiality agreements to the TTO, sponsored programs office, or general counsel rather than individual faculty, precisely because an unreviewed NDA can bind the institution to unfavorable terms without anyone checking them against university policy first. Even a vendor’s short, “standard” NDA is worth routing through the appropriate office.
How long does a research NDA typically last?
The confidentiality obligation itself commonly runs three to seven years from the date of disclosure, independent of how long the underlying discussions or relationship last. Shorter terms are appropriate for less sensitive technical information; longer terms may be justified for trade-secret-level disclosures.
What’s the difference between an NDA and a material transfer agreement?
An NDA protects confidential information — data, descriptions, business terms. An MTA governs the physical transfer of a tangible research material (a cell line, reagent, or compound) and the restrictions on how the recipient may use, store, or redistribute it. A single collaboration can involve both: an NDA to discuss the material’s background and potential uses, and a separate MTA to actually ship it.
Is an NDA required before a sponsored research agreement can be signed?
Not always, but it’s common practice when the parties need to exchange non-public information to scope the collaboration before committing to a funded scope of work. The NDA covers that preliminary conversation; it does not substitute for the SRA, which separately negotiates IP ownership, funding, deliverables, and publication terms for the actual research.
Related CASRAI resources
- Confidentiality agreement (NDA) — the CASRAI Dictionary’s operational definition
- Industry-University Research Partnerships: Agreement Structures and IP Terms
- Material Transfer Agreements (MTA): The Practical Process for Sharing Research Materials
- Sponsored research agreement
- Material Transfer Agreement (MTA)
- Visiting scholar agreement
- Data Use Agreement (DUA)
- Data Sharing Agreement (DSA)
- Export Control (EAR/ITAR) and International Research Collaboration
- Technology Transfer & Innovation — cluster overview
