A new drug candidate has a 90% chance of dying before it reaches a pharmacy shelf.
Most people understand that medicine is hard, but the specific shape of the failure is opaque. The process is not a linear climb toward approval. It is a narrowing funnel where candidates are eliminated at specific gates for specific reasons. The 1 in 10 success rate is not a reflection of bad science, but of rigorous safety thresholds enforced by the U.S. Food and Drug Administration.
To understand why the attrition is so high, you must separate the three distinct goals of the clinical pipeline. Safety, efficacy, and scale are tested sequentially, not simultaneously. A drug that is safe in 20 people is not necessarily safe in 2,000. A drug that lowers blood pressure in a mouse is not necessarily effective in a human. The data tracked by the Biotechnology Innovation Organization and the FDA Center for Drug Evaluation and Research (CDER) shows where the math breaks down.
The starting line is the Investigational New Drug (IND) application. This is the permission to move from animal models to humans. Once the IND is cleared, the drug enters the clinical trial phases. Each phase has a pass rate that compounds the risk.
The funnel, in numbers
The following table breaks down the probability of a drug moving from one stage to the next. These rates are aggregated from FDA CDER annual reviews and industry benchmarks tracked by the Biotechnology Innovation Organization (BIO).
| Phase | Pass Rate | Cumulative Probability | Primary Failure Mode |
|---|---|---|---|
| Phase 1 (Safety) | 63% | 63% | Toxicity or pharmacokinetics |
| Phase 2 (Efficacy) | 31% | 19.5% | Lack of clinical effect |
| Phase 3 (Scale) | 58% | 11.3% | Safety signals at scale |
| FDA Review (BLA) | 85% | 9.6% | Data integrity or labeling |
The math is multiplicative, not additive. A 63% chance of passing Phase 1 does not mean you start Phase 2 with a 37% risk. It means you enter Phase 2 with a 63% chance of being there at all. By the time a drug reaches the FDA for a Biologics License Application (BLA) or New Drug Application (NDA), the cumulative probability of success is roughly 10%.
The steepest drop occurs between Phase 1 and Phase 2. This is the transition from “is it safe?” to “does it work?” Phase 1 trials typically involve 20 to 100 healthy volunteers or patients. The International Council for Harmonisation (ICH) guidelines define this phase primarily by pharmacokinetics—how the body absorbs, distributes, and excretes the drug. If the liver metabolizes the compound too quickly, or if the drug causes an unexpected spike in heart rate, the trial stops.
Phase 2 is where the majority of drugs fail. This phase tests efficacy in 100 to 300 patients with the target disease. Here, the drug must demonstrate a statistically significant improvement over a placebo or standard of care. Many candidates pass safety but fail to show a meaningful biological effect. A tumor might shrink by 5% in a Phase 2 cohort, but if the control group shrinks by 2% due to regression to the mean, the drug offers no clinical advantage.
Phase 3 trials involve 1,000 to 3,000 patients across multiple sites. The pass rate improves to 58% because the drug has already proven it works in a smaller group. However, the sample size is large enough to catch rare adverse events. A side effect that appears in 1 in 500 patients might be invisible in Phase 2 but becomes statistically significant in Phase 3. This is why the FDA requires Phase 3 data: to ensure the risk-benefit profile holds across a diverse population.
The synthesis
The shape of the funnel reveals a specific tradeoff: speed versus certainty. If the FDA lowered the bar for Phase 2 success, more drugs would reach Phase 3, but more would also fail at the scale-up stage, wasting capital and exposing patients to ineffective treatments. The 31% pass rate from Phase 2 to Phase 3 is a feature, not a bug. It ensures that only candidates with robust efficacy signals move to the expensive, large-scale Phase 3 trials.
The FDA review process itself has a high pass rate (85%) for submissions that reach it. This indicates that the elimination happens in the trials, not at the decision desk. By the time a Biologics License Application is submitted, the sponsor has already invested hundreds of millions of dollars and spent 10 years on average to validate the data. The FDA’s role is to verify the data integrity and ensure the labeling matches the evidence.
The cumulative 9.6% success rate is often cited as the “1 in 10” rule. This number is critical for understanding drug pricing. Because 9 out of 10 candidates generate zero revenue, the successful drug must recoup the costs of the 9 failures plus the cost of its own development. This economic reality is not hidden in the science; it is baked into the math of the pipeline.
The closer
The 1 in 10 success rate is not a measure of how many drugs are “good,” but how many survive the specific safety and efficacy gates defined by the ICH and enforced by the FDA. The math says a Phase 2 failure costs the most in wasted time, while a Phase 3 failure costs the most in wasted capital. The 63% Phase 1 pass rate ensures safety first, but the 31% Phase 2 pass rate ensures the drug actually works before the industry bets the house on it. That 90% failure rate is the cost of the guarantee that a prescription on the shelf is both safe and effective.