Three phases, three questions
Clinical development is staged so that each step risks fewer people in exchange for a clearer answer, and only graduates a molecule that passes. The three classic phases each have a distinct job:
- [[phase-i-trial|Phase I]] — is it safe in people? Usually 20–100 healthy volunteers (or patients, for toxic drugs like cancer therapies). The dose is escalated slowly to find what the body tolerates and to measure human PK. This is the realm of the first-in-human study.
- [[phase-ii-trial|Phase II]] — does it work, and at what dose? A few hundred *patients* who actually have the indication. This is the first real test of efficacy and the place to find the dose with the best therapeutic window.
- [[phase-iii-trial|Phase III]] — is it better, in many patients? Hundreds to thousands of patients in large, randomized, controlled trials that confirm benefit, detect rarer side effects, and provide the evidence regulators need.
Why trials grow — and why Phase II kills
The expanding size is deliberate. Early on, you expose as few people as possible to an unknown molecule. Only after Phase I shows tolerability, and Phase II shows a real signal, is it ethical and economical to enroll the large numbers needed to prove benefit statistically and catch rarer harms. Big claims demand big evidence.
What "works" actually means in a trial
A trial does not just observe patients getting better — people get better for many reasons, including the placebo effect and the natural course of illness. To credit the drug, Phase III trials are typically randomized (patients are assigned to drug or comparator by chance), controlled (against placebo or standard of care), and often blinded (neither patient nor doctor knows who got what). The trial is judged on a pre-declared *endpoint* — a specific measure of benefit chosen before the data are seen, so the answer cannot be cherry-picked afterward.