If you have wondered why do drugs fail in clinical trials, the short answer is that among products that failed in the cited pivotal-trial study, most failed because of inadequate efficacy, while some failed because of safety concerns.
The sections below explain the main reasons reported in published research and make no predictions about any individual drug.
| Measure | Reported value |
|---|---|
| Phase II failures attributed to lack of efficacy (most recent year range in the Nature analysis) | 59% |
| Phase III and beyond failures attributed to lack of efficacy (most recent year range in the Nature analysis) | 52% |
| Phase II failures attributed to safety (most recent year range in the Nature analysis) | 22% |
| Phase III and beyond failures attributed to safety (most recent year range in the Nature analysis) | 35% |
| Pivotal-trial failures due to inadequate efficacy | 57% |
| Pivotal-trial failures due to commercial reasons | 22% |
Figures come from Nature Reviews Drug Discovery and a PubMed-indexed JAMA Internal Medicine study. They can change as new analyses appear, so check the sources before you rely on them.
This guide is part of our wider series on what is drug discovery, and it looks at where candidates drop out and why. It is educational only and says nothing about any specific drug’s effects.
How clinical development filters candidates
Each phase of clinical development asks a different question.
- Phase I mainly looks at safety and dosing in a small group.
- Phase II asks whether the candidate shows signs of benefit in people with the condition.
- Phase III confirms benefit and monitors side effects in much larger groups.
Because every gate removes some candidates, only a minority of those that enter human testing reach regulatory review. The exact share varies by era, disease area and data source, so any single percentage is best read as a rough guide.
The JAMA Internal Medicine analysis tracked late-stage programmes only. Its authors concluded that “Roughly half of investigational drugs entering late-stage clinical development fail during or after pivotal clinical trials, primarily because of concerns about safety, efficacy, or both.”
Why do drugs fail in clinical trials? The main reasons
Published analyses sort failures into a few broad causes. Lack of efficacy is reported as the largest cause, while safety problems and commercial or strategic decisions account for much of the rest.
In the most recent year range of the Nature Reviews Drug Discovery analysis, the proportion of failures due to lack of efficacy was higher in Phase II (59%), but “still disturbingly high in Phase III and beyond (52%).”
The JAMA Internal Medicine study looked at products that entered pivotal trials between 1998 and 2008, with follow-up through 2015. It reported that most failed for inadequate efficacy (57%), 17% for safety and 22% for commercial reasons.
Efficacy: the benefit is too small or absent
A candidate may fail because its benefit is inadequate.
Safety: problems that appear at scale
Early studies are small, so rarer side effects may stay hidden. For the most recent year range, the Nature Reviews Drug Discovery authors found safety failures were higher in Phase III and beyond (35%) than in Phase II (22%). They suggest this “may be due to safety issues that only become apparent in larger numbers of patients and/or longer trials.”
Commercial and strategic reasons
Some candidates stop even when no scientific barrier appears. A sponsor may reprioritise its portfolio or judge that the market is too small. The JAMA Internal Medicine study counted these under commercial reasons (22%), and few such trials were later published.
Where in the pipeline the losses concentrate
The cited data show lack of efficacy remaining the leading cause of failure into later phases.
A late failure is the most costly kind. It follows years of work and the largest trials, which is why there is strong interest in finding weak candidates earlier.
Categories overlap, and a drug may stop for efficacy and safety together.
What earlier work can and cannot change
For related reading, see the guide to hit to lead optimization.
Another related guide covers cell imaging in drug discovery.
Computational and AI methods are often discussed as ways to improve these early steps. The published evidence here does not show that they have lowered clinical failure rates, and any such claim needs data from trials that have reported results.
- Models can only learn from the data they are given, and failed programmes are often unpublished.
- Animal and cell systems do not fully reproduce human disease.
- Commercial failures are not mainly a scientific question, so better prediction may not address them.
How to read attrition statistics carefully
Attrition numbers look precise, but they depend on how each study defines success. They also depend on which years, companies and disease areas were included.
- Check the time window, since the mix of reasons has shifted between periods.
- Check whether the figure covers a single phase or the whole path to approval.
- Check how many failures had a known reason.
- Prefer peer-reviewed analyses over summaries that do not cite their data.
Treat these statistics as context for how hard drug development is. They are not a prediction about any single candidate, and they are not medical or investment advice. Anyone with questions about their own care or a specific trial should talk to a physician or another qualified health professional.
It also helps to remember what a “failure” means. Sharing those results is valuable, yet the JAMA Internal Medicine study found that trials of products stopped for commercial reasons were rarely published.
Missing results make it harder for later teams to learn from earlier work. They also make attrition figures less complete than they appear, which is one more reason to read every percentage with some caution.
Finally, the picture changes with the disease area. Conditions with well-understood biology and clear measures of response are generally easier to test than those where the cause is poorly understood. A single headline figure therefore hides large differences between fields, and readers should look for analyses that break results down before drawing conclusions for any one area.
Frequently asked questions
What is the most common reason a drug fails in a trial?
Show answer
Lack of efficacy. In the most recent year range of the Nature Reviews Drug Discovery analysis, it accounted for 59% of Phase II failures and 52% of failures in Phase III and beyond.
Do safety problems cause many failures?
Show answer
Yes, particularly later. Safety accounted for 22% of Phase II failures and 35% of Phase III and beyond in the most recent year range of the Nature Reviews Drug Discovery analysis.
Do drugs ever stop for non-scientific reasons?
Show answer
Yes. The JAMA Internal Medicine study attributed 22% of late-stage failures to commercial reasons.
Can AI stop drugs from failing?
Show answer
The evidence cited here does not establish whether AI can reduce clinical trial failures.
This article is for information only. It is general education about science and the drug industry, and it is not medical, treatment or investment advice.
