Inside a Clinical Trial: What Phase 1, 2, and 3 Results Really Mean

Imagine following one experimental drug through five years of headlines. Year one: “New compound is safe in first human study.” Year three: “Drug shows promising signs in patients.” Year five: “Late-stage trial misses its main goal.” Nobody lied. Each headline described a different experiment asking a different question—and a reader who treated the first two as a verdict would be startled by the third.

That imaginary drug is a composite, but the pattern is real, and it is why the word “phase” carries so much weight in medical news. Knowing what Phase 1, 2 and 3 are actually designed to show is the single most useful skill for reading a biotech story honestly.

Why human testing comes in stages

Before any person receives an experimental drug, it goes through preclinical work—experiments in cells and animals. In the United States, a company or research group then files an Investigational New Drug application (IND), the paperwork that asks the Food and Drug Administration for permission to start testing in people.

From there, a clinical trial—a study in which volunteers receive an intervention so researchers can measure its effects—proceeds in phases. Each phase is a gate. The FDA’s own patient guide to clinical research sketches the typical shape:

• Phase 1: roughly 20 to 100 healthy volunteers or people with the condition, lasting several months, focused on safety and dosage.
• Phase 2: up to several hundred people with the condition, lasting several months to two years, looking at efficacy and side effects.
• Phase 3: roughly 300 to 3,000 volunteers with the condition, lasting one to four years, designed to demonstrate benefit and monitor adverse reactions.
• Phase 4: studies after approval, often involving several thousand people, watching safety and effectiveness in wider use.

The same FDA page estimates that about 70% of drugs move from Phase 1 to the next phase, about a third move on from Phase 2, and 25–30% move on from Phase 3. Other sources use slightly different ranges—the NIH’s introduction to clinical research, for example, describes larger Phase 3 studies—because these are typical shapes, not rules. The direction of travel is the point: most candidates that enter human testing never reach patients.

Phase 1: Is it safe enough, and at what dose?

A Phase 1 trial is mainly a safety and dosing study. Researchers start with low doses and cautiously step up, watching for side effects and tracking how the body absorbs, processes and clears the drug. The National Cancer Institute describes Phase 1 as the stage where researchers figure out whether a new treatment is safe and what its side effects are.

In many fields, Phase 1 volunteers are healthy. In cancer, they are often patients who have run out of standard options, because giving a potentially toxic drug to healthy people would not be ethical. That is why cancer Phase 1 studies sometimes report tumours shrinking in a handful of people—an intriguing hint, but one drawn from small groups, usually without a comparison group, and not designed to prove benefit.

Plain-English: An endpoint is the specific thing a trial measures to decide whether a treatment worked—for example, how long people live, whether a tumour shrinks, or whether a blood marker changes.

Phase 2: Hunting for a signal

Phase 2 asks whether the drug seems to do something useful in people with the condition, and at which dose. Trials are bigger, but often still small enough that chance can flatter a result. Many rely on surrogate endpoints—markers such as a lab value or a scan result that are hoped to predict real benefit but are not themselves a direct measure of how patients feel, function or survive.

This is the phase where hope most often outruns evidence. In a report on 22 case studies where Phase 2 and Phase 3 trials had divergent results, the FDA described products that looked promising in Phase 2 and then disappointed in larger trials: 14 on efficacy alone, one on safety alone, and seven on both. The agency noted that surprises happened even when Phase 2 studies were relatively large and even when they measured clinical outcomes.

Phase 3: The confirmatory test

Phase 3 is where a treatment has to prove itself against something. The key design features are worth knowing by name:

• Randomization: participants are assigned to groups by chance, which the NCI notes helps prevent bias.
• A control group: people receiving the current standard treatment, a placebo, or both. A placebo is an inactive product that resembles the test product. The NCI points out that placebos are rarely used in cancer treatment trials; new cancer drugs are usually compared with standard therapy.
• Blinding: in a double-blind trial, neither participants nor the research team know who got what, so expectations cannot colour the results.
• A pre-specified primary endpoint: the main question is fixed before the data come in, so researchers cannot pick the most flattering result afterwards.

When a Phase 3 trial “meets its primary endpoint,” it means the difference on that main measure was unlikely to be due to chance by the trial’s pre-set statistical standard. It does not automatically mean the difference is large, that every patient benefited, or that side effects are acceptable. Statistically significant and clinically meaningful are two different questions.

A trial phase is not a grade for a drug. It is a description of which question has been asked—and which ones have not been asked yet.

A short decoder for trial-speak

• “Topline results”: a company’s first summary of the headline numbers, usually before full peer-reviewed data are public.
• “Well tolerated”: a judgement, not a measurement. Look for the actual rates of serious side effects and of people stopping treatment.
• “Numerically improved”: the number went the right way, but not convincingly enough to rule out chance.
• “Open-label”: everyone knows who is getting the drug, which makes some outcomes easier to over-read.
• “Subgroup”: a slice of trial participants. Findings confined to a subgroup, especially one not planned in advance, usually need their own confirmation.

After Phase 3: Approval is a decision, not the finish line

Positive trials lead to a marketing application, and regulators weigh the whole package: efficacy, safety, manufacturing quality and more. For serious conditions with unmet needs, the FDA has since 1992 been able to grant accelerated approval based on a surrogate endpoint that is “reasonably likely” to predict clinical benefit—tumour shrinkage in cancer is the classic example. The trade-off is explicit: the company must run confirmatory trials, and the FDA says approval may be withdrawn or the label changed if those trials fail to verify clinical benefit.

The FDA also publishes a table of surrogate endpoints that have supported approvals, separating markers “known to predict” benefit from those only “reasonably likely” to. Whether a given marker is acceptable is decided case by case. And after approval, Phase 4 and ongoing safety monitoring continue, because rare side effects may only appear once many more people use a medicine.

Six questions for the next trial headline

1. Which phase is it? Phase 1 news is about safety and dose, not proof of benefit.
2. How many people, and was there a comparison group? Small, uncontrolled studies generate hypotheses rather than answers.
3. Was it randomized and blinded? If not, expectations and selection can shape the result.
4. What was the primary endpoint—and did the trial meet it? Be wary when a release leads with secondary or subgroup findings.
5. Is the endpoint a surrogate or a direct measure of how patients live, feel or function?
6. What happened to safety? Look for serious adverse events and dropouts, not just the word “tolerated.”

None of this makes early trials unimportant. Every approved medicine passed through a Phase 1 that reported “safe enough to continue” and a Phase 2 that found a signal. The point is to give each result the weight its design can bear. For more on reading early science, see our guide to how to read a biotech breakthrough, the Breakthrough Reality Check, and our look at what happens between a promising molecule and a medicine. Or walk the whole route yourself in the Drug Development Journey.