Somewhere in the world, about every 12 minutes, a new clinical trial begins. That is not a slogan — it is the measured pace of registered clinical research in 2026. This study queries every study in the ClinicalTrials.gov registry with a start date between January 1 and October 14, 2026: 33,678 trials. Every number below comes from the registry's own public API, queried on October 15, 2026, and every query is documented in the methodology so anyone can reproduce it.

What follows is a census of the global clinical-research engine: how fast it runs, what it tests, who pays for it, and where it happens.

Key findings

  • 33,678 studies started in 288 days (January 1 to October 14, 2026) — roughly one new trial every 12 minutes, around the clock.
  • The engine is running about 6% hotter than 2025: 116.9 trial starts per day so far in 2026 versus 110.1 per day last year. At this pace the full year lands near 42,700 registered studies.
  • The late-stage pipeline is accelerating faster than the total: annualized 2026 phase-3 starts (~2,343) are up about 15% over all of 2025 (2,042). Industry leads 53% of phase-3 trials, but only 17.5% of all trials.
  • Cancer dominates: 4,463 trials list a neoplasm condition — 13.3% of everything registered. Diabetes (847), breast cancer (818), and obesity (783) follow. COVID-19, once the registry's biggest category, is down to 62 new starts.
  • China is closing on the United States: 4,896 trials have at least one Chinese site versus 7,675 with at least one US site — China now runs nearly two-thirds as many as the US.
  • Academia, not pharma, starts most trials: 79.1% of lead sponsors are academic institutions, hospitals, or other non-industry organizations. The NIH itself is the lead sponsor on just 110 starts (0.3%).
  • 40.3% of the year's starts are recruiting right now (13,591 studies), and 16.3% include children (5,495 studies).

Journalists are welcome to quote any finding above with credit to Calder Brief. Underlying counts were pulled from ClinicalTrials.gov's public API v2 on October 15, 2026; see Methodology for exact queries.

The cadence: a trial every 12 minutes

The registry logged 33,678 studies with start dates in the first 288 days of 2026 — 116.9 per day, or one every 12.3 minutes. For comparison, the full year 2025 produced 40,202 starts (110.1 per day). Month by month, the pace looks like this:

October covers only the 1st through the 14th. September's high (4,275) partly reflects registration backfill — studies are often registered after they start, so earlier months are slightly undercounted relative to later ones.

What they're testing

Interventional studies — the ones that actually test a treatment, device, or procedure — outnumber observational studies nearly four to one:

26,468 interventional (78.6%) vs 7,210 observational (21.4%) of 33,678 total.

The phase funnel

Of the 26,468 interventional trials, phase 2 is the busiest stage and phase 4 (post-marketing surveillance) the quietest:

Phase labels overlap: many trials carry multiple phase designations, and 16,663 interventional studies have no phase recorded (devices, behavioral interventions, diagnostics).

The striking number is at the end of the funnel. There were 1,849 phase-3 starts in 288 days — an annualized ~2,343, up about 15% over all of 2025 (2,042). The late-stage pipeline is growing faster than the registry as a whole, and it is where industry concentrates: industry is the lead sponsor on 987 of the 1,849 phase-3 trials (53.4%), versus 807 led by academic or other non-industry sponsors (43.6%), with the remainder led by government sponsors.

Who pays for the science

Lead-sponsor class, across all 33,678 studies:

The popular image of clinical research — pharma companies running trials — describes only the expensive end of the funnel. Four out of five registered studies are led by academic institutions, hospitals, and other non-industry organizations (26,634, or 79.1%). Industry leads 5,885 (17.5%), other government agencies lead 1,049 (3.1%), and the NIH itself leads just 110 (0.3%) — it far more often collaborates than leads. Industry's influence is concentrated where the costs are: those 17.5% of starts include most of the phase-3 work that brings drugs to market.

What diseases get studied

Trials by condition (overlapping categories — one trial can list several conditions):

Cancer is the runaway leader: 4,463 trials list a neoplasm condition, 13.3% of everything registered. Then a second tier of chronic-disease research — diabetes (847), obesity (783), depression (570), hypertension (451) — reflects where the treatment markets and public-health burdens are. The most telling number may be the smallest: COVID-19 drew just 62 new trial starts, a category that once dominated the registry now reduced to a rounding error.

Where the research happens

Trials counted by country of at least one trial site (multinational trials count in every country they operate in):

The United States remains the center of gravity at 7,675 trials with at least one US site. But China, at 4,896, now runs nearly two-thirds as many — a gap that would have been unthinkable a decade ago. France (1,785), Canada (1,027), and the United Kingdom (862) round out the next tier, with Germany, Australia, Japan, and India behind them. The geography of clinical research is still Western-led, but the center of mass is shifting east.

Methodology

All counts come from ClinicalTrials.gov, the US National Library of Medicine's registry, via its public API v2 (https://clinicaltrials.gov/api/v2/studies), queried on October 15, 2026, using the countTotal=true parameter with pageSize=1 — i.e., the counts are the registry's own totals, not scraped or estimated.

  • Population: studies with AREA[StartDate]RANGE[01/01/2026,10/14/2026], the study-start date as registered. 2025 comparison: RANGE[01/01/2025,12/31/2025].
  • Monthly trend: one query per month with that month's date range (October: 10/01/2026–10/14/2026).
  • Type: AREA[StudyType]INTERVENTIONAL / OBSERVATIONAL.
  • Phase: AREA[Phase] values PHASE1–PHASE4, EARLY_PHASE1 on the interventional base. The registry's query enum does not expose combined phases (Phase 1/2, Phase 2/3) as separate values, so phase labels overlap and do not sum to the interventional total.
  • Sponsor: AREA[LeadSponsorClass] INDUSTRY / NIH / OTHER_GOV / OTHER (OTHER is dominated by academic institutions and hospitals).
  • Conditions: AREA[Condition] term matching; categories overlap.
  • Countries: AREA[LocationCountry]; a trial with sites in several countries is counted in each, so country counts overlap.
  • Status / age: AREA[OverallStatus]RECRUITING; AREA[StdAge]CHILD.

Caveats: registration lags reality — studies are often registered after they start, so recent months (especially October, which is partial) are undercounted relative to earlier ones; September's high partly reflects backfill. “Start date” is the registered date, which sponsors can revise. Phase and country totals intentionally double-count by design of the registry's fields. Annualized projections divide 2026 counts by 288 days and multiply by 365. All queries were re-run on October 15, 2026, and the numbers above are exactly what the API returned.

Sources

Corrections policy: if you spot an error in the data or its interpretation, tell us and we will fix it and note the change here.