Multi-PI Grant Proposals: Leadership Plans, Budgets, and Making Collaboration Work

A multi-PI grant is not the same thing as a grant with multiple co-investigators. It is a different structural beast — one with its own governance requirements, its own review criteria, and its own high-failure modes that have nothing to do with the science. The critical difference lies in how NIH evaluates what is called the Multiple PI Leadership Plan, and when that evaluation goes wrong, it can sink an application even if your Specific Aims are exceptional.

On first reading, the leadership plan looks like an administrative checkbox. Five to ten pages of boilerplate about meetings, decision-making, and how you will communicate. In practice, it is the single highest-value real estate in a multi-PI proposal, because it sits under review Factor 3 (Expertise and Resources) and Factor 3 is now evaluated as Pass/Fail, not scored. That means a "Not Sufficient" rating on governance can torpedo your entire application regardless of scientific merit. This post walks through what that structure means, where proposals fail, and how the budget allocation of your grant is a hidden signal to reviewers about whether your collaboration is genuine or ceremonial.

When multi-PI makes sense, and when it doesn't

Start by asking whether you actually need a multi-PI structure. A multi-PI grant designates two or more Principal Investigator equivalents (PD/PIs) — each with independent authority, each with a separate biosketch, each earning their own percent effort commitment in the budget. This is not the same as a single-PI grant with multiple co-investigators (where the lead PI retains scientific and administrative authority and co-invs serve in a supporting role).

Multi-PI is the right structure when:

• You have genuinely equal scientific contributions. Not "co-authors" — equal responsibility for the direction of the work. If one PI is substantially leading and others are supporting, that is a co-investigator structure, and it is simpler.
• The project spans multiple independent research programs. A plant biology lab and a statistical genetics lab both have intellectual ownership of different research domains in the same grant. Each brings non-overlapping expertise that cannot be easily compressed into one lab.
• The PIs are at different institutions. Multi-institutional projects can use a single-PI structure (the lead institution applies, subawards go to collaborators), but multi-PI is clearer when the institutions are truly equal partners in the intellectual direction.

Multi-PI is not the right structure when:

• You are using it to give a senior collaborator a ceremonial co-PI title. Reviewers are trained to spot this. If the budget shows 90 percent at one site and 10 percent at another, that second person is not a co-PI — they are a consultant with a better title.
• You want to work around the single-PI percent-effort ceiling. Yes, NIH caps single-PI effort at a certain threshold depending on the mechanism, but a multi-PI structure that exists only to double that ceiling is administratively transparent and will be flagged by a program officer as a policy violation.
• One PI will functionally make all major decisions. If that is the case, make it honest. Single PI with strong co-invs and an advisory board is more defensible than a multi-PI structure where one person holds de facto veto power.

The Multiple PI Leadership Plan: structure and what reviewers look for

The NIH guidance is formally titled "Guidance for Multiple Principle [sic] Investigator/Project Director Submissions" and it sits in the NIH Grants Policy Statement. The required elements are:

1. A clear description of the roles and responsibilities of each PI. Not job titles — actual responsibilities. What does PI #1 own? What does PI #2 own? Where do they overlap?
2. A description of how the PIs will make decisions, especially when they disagree. Do you vote? Does one person break ties? Is there an external arbiter? The absence of a clear protocol is a red flag.
3. A conflict-resolution mechanism. Explicitly name what happens if a PI wants to take the project in a direction another PI opposes.
4. A governance structure that ensures sustained collaboration. How often do you meet? Who chairs meetings? Is there a steering committee? An external advisory board? How often do they convene?
5. A plan for what happens if a PI departs. Voluntary departure, illness, or job change — how is the work absorbed?

The second-most common failure mode is a plan that is technically complete but reads like it was imported from a template. Reviewers see dozens of applications that say "PIs will meet monthly and communicate via email." That is not a plan; that is a bare minimum. A strong plan is specific about your actual collaboration, references the actual labs and the actual scientific problem, and demonstrates that you have thought about the coordination overhead.

Here is the honest uncomfortable thing: running a multi-PI project has real coordination costs. The best proposals name this explicitly. "Our labs are 40 minutes apart, which allows in-person meetings but requires structured asynchronous reporting. We budget two weeks of graduate student time per year for data harmonisation between sites." That sentence does more for credibility than ten sentences of generic governance language.

Budget allocation as a signal of genuine collaboration

One of the most underrated signals in a multi-PI proposal is where the money actually goes. Reviewers look at the budget allocation across sites and ask a very direct question: if this were just a single-PI grant with a consultant, would the budget distribution still make sense?

A budget that puts 85 percent of funds at Site A and 15 percent at Site B tells reviewers something specific. If you say "Site B is the data-coordinating centre and does statistical analysis," then 15 percent might be defensible. If you say "Site B brings equal expertise in immunology," then 15 percent is indefensible — it signals Site B is not actually doing equal work, and the application lacks internal honesty.

The gold standard for a genuine multi-PI collaboration is a budget that reflects the intellectual division of labour. If PI #1 leads the wet-lab work and PI #2 leads the modelling, the budget distribution should reflect that split. Not necessarily 50-50 (one component might be more expensive), but in proportion to the actual effort and intellectual contribution. Reviewers can read a budget.

Here is a concrete example. Suppose you have a consortium grant with PI #1 at a research-1 university (F&A rate 26 percent) and PI #2 at a smaller institution (F&A rate 18 percent). Your direct costs are $600,000. If you allocate $400,000 to Site A and $200,000 to Site B:

What matters is that the allocation is defensible and consistent with the scientific narrative. If your Research Strategy says "the project is divided equally in intellectual contribution but Site B handles data management and Site A handles experimental work," then you need the budget to match. If it doesn't, reviewers will notice, and they will interpret the mismatch as evidence that you either haven't thought carefully about the division of labour or aren't being honest about it.

Letters of support: specific versus generic

A generic letter of support might read: "Dr. X is an outstanding collaborator with deep expertise in molecular biology. We are confident in the success of this collaboration and look forward to supporting this important work."

Reviewers interpret that letter as: "The collaborator has not read this proposal."

A strong letter references specific aims by name, describes what the collaborator will actually contribute, and shows evidence of deep reading:

In Aim 2, which relies on structural predictions from molecular dynamics, I will conduct all MD simulations using the GROMACS pipeline that I developed for our earlier work on protease dynamics (Chen et al., 2024). Dr. X brings no direct expertise in this step. Rather, I will rely on Dr. Y's protein chemistry background to validate the structures against biochemical data from our pre-existing collaboration. The iterative feedback between simulation and experiment is where my expertise adds the most value. My lab will conduct the MD work and contribute equipment (dual-GPU workstations); Dr. Y's lab will conduct all wet-lab validation and contribute reagents and cell lines valued at approximately $45,000 over the project period.

Notice what that letter does: it names the collaborator, it references a specific aim, it describes the actual intellectual contribution, it acknowledges what the collaborator is not doing, and it itemises the concrete resources. It is specific enough that a reviewer could read that letter and reconstruct why the collaboration makes sense.

Your letters should be substantive. One page minimum. Include them as appendix material. They do not count against your page limit, so there is no excuse for thinness.

Coordination overhead: the hidden budget item

One of the biggest risks in multi-PI grants is not scientific disagreement — it is the coordination overhead that eats into research time and does not appear in the budget because it is invisible.

Good proposals anticipate this. You might budget a graduate student half-time for data harmonisation, or allocate a senior staff member 5 percent effort to serve as a coordination hub. You explain why in the budget justification. "Site A and Site B use different sequencing pipelines; weekly harmonisation calls and quarterly data reviews require structured coordination."

The worst proposals pretend coordination is free. They assume monthly Zoom calls will magically keep a multi-site project aligned. In reality, coordinating across sites means duplicate lab meetings, asynchronous documentation, delayed decision-making, and the occasional discovered incompatibility between datasets that should have been caught earlier. Name it. Budget for it. Reviewers know it exists, and naming it is a sign of realistic planning.

Governance structures that actually work

There are three components to good governance in a multi-PI grant:

1. Decision-making protocol

Define who decides what. Some decisions are scientific (which analyses to prioritise, how to handle unexpected findings). Some are administrative (hiring, reallocation of budget between sites). Some are collaborative (timeline adjustments, publication decisions). You need different protocols for each.

Common protocol structures:

• Consensus with escalation. PIs try to agree on all decisions. If consensus fails on a scientific matter, the issue goes to the external advisory board. If consensus fails on an administrative matter, the Contact PI decides.
• Tiered voting. Decisions affecting the entire project require all PIs. Decisions affecting one site require that site's PI. Tied votes go to an independent arbiter (advisory board chair, program officer).
• PI-of-record for each aim. Aim 1 is led by PI #1 (final decision authority), Aim 2 by PI #2, with explicit hand-off protocols where aims intersect.

The specific protocol matters less than clarity. A reviewer wants to be able to read your document and know exactly who decides what.

2. External advisory board

Naming an external advisory board in your proposal is a signal that you have thought about governance beyond "we'll meet monthly." The board should include people outside your collaboration who can offer perspective on the science and help adjudicate disagreements if they arise.

Two or three external advisors is typical. They should have complementary expertise — if you have a data person and a biology person on the PIs, get someone with a perspective from the third corner (e.g., clinical translation, industry application, methodological rigour). Name them in the proposal and get their cv/biosketch as appendix material. NIH allows this and it strengthens the application.

3. Steering committee and annual meetings

Define a steering committee (probably the PIs plus external advisors). Commit to annual in-person meetings and quarterly teleconferences. In the budget justification, allocate modest travel funding for these — it is a legitimate expense and it signals commitment. "We budget $2,000 per PI per year for annual steering committee meetings. These in-person meetings are critical for aligning long-term direction and reviewing progress against milestones."

Multi-institutional coordination: IRBs, contracts, data sharing

When your PIs are at different institutions, there are mechanical coordination challenges that must be addressed in the proposal:

IRB and regulatory approvals

Each institution needs its own IRB approval (or IBC approval for recombinant DNA, IACUC for animal work). This is not a paperwork issue — it is a scientific issue. Your timelines must account for the longest approval cycle. Some institutions are fast; others take months. Build that into your project timeline.

In your Leadership Plan, specify that "IRB approvals will be coordinated through the applicant institution's compliance office, with each site's IRB conducting independent review. The project will not begin data collection until all institutional approvals are in place." That sentence signals you understand the constraint.

Subcontract and consortium agreements

When one institution is the lead applicant (the one submitting to NIH) and the others are subawardees, there is a subcontract or consortium agreement. This is usually handled by grants offices and you do not need to draft it yourself, but you should know it exists and allow time for execution. Some institutions take weeks to process subawards.

A practical note: start the subcontract paperwork before submitting the application. If you wait until after the award, you may discover that one institution's conflict-of-interest policy or IPR requirements are incompatible with the project. Better to find that out early.

Data sharing and intellectual property

Specify in the Leadership Plan how data will be shared between sites. Real answer, not generic language. "Site A will conduct RNA-seq; raw reads and processed count tables will be shared with Site B within two weeks of completion. Site B will conduct all downstream statistical analysis; code and results will be shared with Site A in real time via a GitHub repository." That is a plan. "Data will be shared regularly and in accordance with NIH policy" is not.

Address IP upfront too. If your collaboration generates a new tool or method, how will authorship and patent applications be handled? The absence of a plan is not a red flag to reviewers until a dispute arises — and then it is a catastrophe. A single sentence in your Leadership Plan preempts future conflict: "Authorship and IP decisions will be made by consensus among all PIs, with disputes adjudicated by the external advisory board."

Roles and responsibilities: Contact PI, PIs, co-investigators, consultants

Get the hierarchy right, because it matters:

• Contact PI (or Responsible Official). This is the PI at the applicant institution (usually the lead institution). All official correspondence with NIH goes through them. The Contact PI is an administrative role, not necessarily the scientific leader. Your team may decide that PI #2 leads the science but PI #1 handles NIH paperwork.
• PD/PI (Principal Director or Principal Investigator). In a multi-PI grant, all PIs are equal in administrative status. Each has their own biosketch, percent effort, and independent authority. Do not use "Co-PI" terminology in formal documents — NIH now uses PD/PI for all principal-level investigators in a multi-PI structure.
• Co-investigator. Lower administrative tier than PI. No formal authority, no percent effort commitment (they can contribute time, but it is not budget-critical). Use this for people who contribute to the science but are not co-equal principals.
• Consultant. External to your institution, providing advice or conducting defined pieces of work. May be paid for their time (through consulting fees) or unpaid. Consultants have no formal authority and are not part of the governance structure.

Use these categories correctly. If someone is genuinely co-equal, make them a PI. If they are supporting one PI's work, make them a co-investigator. The wrong categorisation confuses reviewers about the actual project structure.

What happens if a PI leaves?

Include a succession plan. You need not predict the future perfectly, but reviewers want to know that you have thought about continuity. A succession plan might read:

If PI #1 becomes unable to continue, PI #2 will assume all leadership responsibilities for Aims 1 and 2 (currently led by PI #1). Dr. Z, who is a senior postdoc in PI #1's lab with 8 years of experience in the relevant methods, will transition to co-investigator status and assume day-to-day responsibility for the experimental work. The scope of the project will not change, but the timeline for Aims 1 and 2 will be extended by 6 months to account for the transition. This plan will be formalised in a Change of PI request to NIH within 30 days of any leadership transition.

That is enough. You are not predicting who will leave or when. You are signalling that you have contingency thinking.

The checklist for multi-PI submissions

Before you submit a multi-PI proposal, run through that checklist. The items are not arbitrary — each one addresses something a reviewer is actively looking for.

Where the science meets the structure

The final piece: your Research Strategy must reflect the multi-PI structure. If your Specific Aims and Methods are written as though one person is doing the entire project, reviewers will question why you need multiple PIs in the first place.

A strong multi-PI research strategy reads as follows: Aim 1 (led by PI #1) does X using method A. Aim 2 (led by PI #2) does Y using method B. These aims are interdependent because Z. The collaboration works because PI #1's expertise in A and PI #2's expertise in B are complementary; neither PI has both capabilities, and the project cannot succeed with just one of them.

If instead you write "We will conduct experiments (using method A, led by PI #1) and analyse the data (using method B, led by PI #2)," reviewers read that as "PI #2 is a bioinformatician we added to appease a program officer." That may not be true, but that is what the narrative conveys.