How Big Is a Funded Horizon Europe Consortium? Data from 20,303 Projects

The Short Answer

We analysed the consortium composition of 20,303 funded Horizon Europe projects in CORDIS. There is no single answer to "how many partners should we have" — but there is a very clear answer per funding scheme, and the spread is narrower than most applicants expect.

Funding scheme	Projects	Median partners	Typical range (p25–p75)	Median countries
RIA — Research and Innovation Action	2,408	14	10 – 18	8
IA — Innovation Action	1,042	17	13 – 23	8
CSA — Coordination and Support Action	1,412	6	2 – 12	4
MSCA Doctoral Network	507	18	14 – 22	8
EIC Pathfinder / Transition	646	6	4 – 7	4
Joint Undertaking IA	185	20	13 – 32	8

If you are building a RIA and your consortium has 6 partners or 30, you are outside the range where 50% of funded projects sit. That is not disqualifying — but it needs to be a deliberate choice you can defend, not an accident of who replied to your emails.

The Statistic You Should Ignore

Across all 20,303 Horizon Europe projects with recorded participants, the median consortium size is 2 partners.

That number is true and completely useless.

Horizon Europe is numerically dominated by single-beneficiary schemes. Of those 20,303 projects, 9,757 — 48% — have exactly one participating organisation. They are ERC grants, MSCA Postdoctoral Fellowships, ERC Proof of Concept projects and EIC Accelerator awards. Every one is a real funded project, and none of them tells a collaborative applicant anything.

Consortium size	Projects	Share
1 organisation	9,757	48.1%
2–4 organisations	3,956	19.5%
5–15 organisations	3,989	19.6%
More than 15	2,601	12.8%

This is why every figure in this article is reported per funding scheme. A programme-wide average of 6.0 partners describes no actual project type.

---

Consortium Size by Scheme

Distinct participating organisations per project, deduplicated, Horizon Europe.

Scheme	Projects	Mean	p25	Median	p75	Max
Joint Undertaking IA	185	25.4	13	20	32	96
MSCA Doctoral Network	507	18.2	14	18	22	41
IA	1,042	18.7	13	17	23	82
RIA	2,408	15.0	10	14	18	159
Joint Undertaking RIA	470	15.1	8	12	17	80
MSCA Staff Exchanges	320	12.2	9	11	15	34
CSA	1,412	8.3	2	6	12	88
EIC Pathfinder / Transition	646	5.5	4	6	7	16
ERC Synergy	177	4.1	3	4	5	12
MSCA PF (Global Fellowship)	634	2.5	2	2	3	7
MSCA PF (European Fellowship)	5,649	1.4	1	1	2	6
ERC	4,705	1.2	1	1	1	8
ERC Proof of Concept	1,065	1.1	1	1	1	3
EIC Accelerator (blended)	514	1.1	1	1	1	4

---

Four Things This Data Shows

1. Innovation Actions are bigger than RIAs — consistently

Median IA is 17 partners against 14 for RIA, and the whole distribution shifts: IA's p25 of 13 is nearly RIA's median.

This is consistent with what IAs do. Demonstration and piloting require end users, technology providers and often regulatory or standardisation partners that a research-stage project does not need. Combined with the budget benchmarks — median IA of €6.00M against €4.97M for RIA — the picture is coherent: more partners, more money, roughly the same funding per partner.

2. CSA has the widest spread in the programme

CSA runs from a p25 of 2 partners to a p75 of 12 — the only scheme where the interquartile range spans a factor of six.

That is because "Coordination and Support Action" covers genuinely different things. A CSA can be a small study run by two organisations or a pan-European networking action with 30 members. There is no CSA norm to conform to; the consortium follows the activity.

3. The outliers are real, and they are enormous

The largest RIA consortium in the data has 159 partners. The largest Joint Undertaking IA has 96. The largest CSA has 88.

These are not data errors — they are genuine large-scale infrastructure and partnership projects. They matter for one practical reason: they drag the mean above the median in almost every collaborative scheme. RIA's mean is 15.0 against a median of 14; Joint Undertaking IA's mean is 25.4 against a median of 20. When someone quotes an "average" consortium size, check which average they mean.

4. Geographic spread is tighter than partner count

Median countries represented, for the collaborative schemes:

Scheme	Median countries	p90	Max
RIA	8	12	38
IA	8	12	20
MSCA Doctoral Network	8	11	17
Joint Undertaking IA	8	12	17
Joint Undertaking RIA	7	12	22
CSA	4	12	35
EIC Pathfinder / Transition	4	6	11

A typical 14-partner RIA draws from 8 countries — so roughly two partners per country on average. Consortia do not spread one partner per country; they cluster, usually because a country contributes both a research organisation and an industrial or end-user partner.

This connects directly to how ties are broken at the funding line. Where proposals tie on score, the General Annexes rank them by call coverage, then Excellence and Impact, then gender balance among lead researchers, and then geographical diversity — specifically, the number of Member States or Associated Countries represented that are not already receiving funds from projects higher up the ranking list. A partner from an under-represented country is not decoration; in a tie it is a ranking factor. See our guide to reading your ESR and resubmitting.

---

Where Funded Partners Actually Come From

Number of Horizon Europe projects with at least one participant from each country:

Country	Projects		Country	Projects
Germany	7,169		Denmark	2,208
Spain	6,097		Portugal	2,130
Italy	5,714		Switzerland	2,121
France	5,447		Finland	1,779
Netherlands	4,501		Norway	1,720
United Kingdom	4,095		Ireland	1,632
Belgium	4,001		Poland	1,538
Greece	2,449		Czechia	1,212
Sweden	2,370		United States	1,132
Austria	2,302		Israel	861

Two observations worth drawing out.

Spain and Italy outrank France. Spanish organisations appear in more Horizon Europe projects than French ones, and Italian organisations very nearly do. The intuition that the programme is dominated by the largest economies does not survive contact with the data — participation intensity is not proportional to GDP.

The Netherlands and Belgium punch far above their size. Both appear in more projects than Sweden, Austria and Denmark combined in the first case, and both sit ahead of every Nordic country. Small, research-intensive, highly networked systems do disproportionately well in a programme that rewards collaboration.

For consortium building, this table is a availability map. Partners in Germany, Spain, Italy and the Netherlands are experienced and abundant. Partners in countries lower down the list are scarcer — and, when a tie is broken on geographical diversity, more valuable.

---

How to Use These Numbers

Match the median for your scheme, then justify deviation. 14 for RIA, 17 for IA, 6 for CSA and EIC, 18 for MSCA Doctoral Networks.

Being small is a capability problem; being large is a management problem. A 6-partner RIA will be asked whether it can deliver the work plan. A 30-partner RIA will be asked whether it can be coordinated. Both questions are scored under Quality and efficiency of the implementation.

Count countries, not just partners. Eight is the collaborative norm. Fewer than four in a RIA or IA is unusual enough to invite comment.

Every partner must do real work. Adding a partner for geographic coverage without a genuine role is visible in the work package tables and in the budget split. At roughly €357k EU contribution per partner in a RIA, a partner with no substantive role is a visible anomaly.

Check who has already been funded in your area. Consortium composition is one of the things evaluators assess against the state of the art. You can search funded projects and their full partner lists in the Project Explorer, or directly on CORDIS.

---

Frequently Asked Questions

How many partners should a Horizon Europe RIA have?

The median funded Research and Innovation Action has 14 participating organisations, with half of all funded RIAs falling between 10 and 18 partners, drawn from a median of 8 countries.

How many partners does an Innovation Action need?

The median funded Innovation Action has 17 partners, with an interquartile range of 13 to 23 — consistently larger than RIA, reflecting the end users, technology providers and standardisation partners that demonstration activities require.

What is the minimum consortium size in Horizon Europe?

Eligibility minimums are set in the call conditions, not by these statistics — collaborative topics typically require at least three independent legal entities from three different Member States or Associated Countries. The figures here describe what funded consortia actually looked like, which is consistently much larger than the eligibility floor.

How many countries should be represented in our consortium?

Median funded RIA, IA, MSCA Doctoral Network and Joint Undertaking IA consortia all span 8 countries. CSA and EIC Pathfinder projects typically span 4.

Does adding more partners improve our chances?

Nothing in this data supports that, and it cannot — CORDIS records funded projects only, with no information about unsuccessful proposals. What the data shows is what funded consortia looked like. Where proposals tie on score, geographical diversity is a formal ranking factor, but that is a tie-break, not a general advantage.

What is the largest Horizon Europe consortium?

The largest RIA in the data has 159 participating organisations. Joint Undertaking IAs reach 96 and CSAs reach 88. These are large-scale infrastructure and partnership projects, and they pull the mean consortium size above the median in nearly every collaborative scheme.

---

Methodology

Figures derived from the 56,589 Horizon Europe and H2020 projects recorded in CORDIS as of August 2026. All figures in this article refer to the 21,200 Horizon Europe projects only; H2020 and earlier framework programmes are excluded. Consortium statistics cover the 20,303 Horizon Europe projects with at least one recorded participating organisation.

Organisation counts are deduplicated per project on the organisation identifier, so an organisation participating in a project is counted exactly once regardless of how many times its participation is recorded. Country counts are distinct countries among the deduplicated participants. Country-level project counts denote the number of projects with at least one participant from that country, so a project spanning eight countries is counted in all eight. Schemes with fewer than 150 funded projects are omitted, as small samples produce unstable percentiles.

CORDIS records funded projects only. It contains no data on unsuccessful proposals, so no success rates are inferred here.

Last updated: August 2026.