Why Does Developing a New Medicine Take Years and Why Do Most Candidates Fail?

The long path is not one test repeated four times. It is a series of different questions. Each question removes candidates that cannot answer it.

Why does it take 10 to 15 years to develop a new medicine?

It takes 10 to 15 years because researchers must build evidence in stages, and the human trials alone usually take several years. A peer-reviewed review in Acta Pharmaceutica Sinica B reports that discovery and development exceed 10 to 15 years for a new drug.

The clock starts before anyone receives the medicine. Researchers first need a candidate worth studying. They then test it outside people, ask the FDA for permission to begin human studies, run trials through Phases 1, 2, and 3, prepare a full New Drug Application, and wait for review.

Each stage answers a narrower question. Does it cause serious harm? Does the body handle it in a usable way? Does it help patients? Does the benefit hold across a larger group, at the proposed doses? A candidate that fails one answer does not become an approved medicine.

The final review has its own weight. According to the FDA's drug-review guidance, an application includes the record from preclinical work through Phase 3, analyses, safety updates, proposed labeling, patent information, and manufacturing information. FDA medical officers, statisticians, and pharmacologists review that record, and inspectors examine trial sites for signs of fabricated, manipulated, or withheld data.

After FDA accepts a complete application, its decision window is 6 to 10 months. That is shorter than the trial years that came before it. Review is the gate. Evidence is the road.

What happens before a medicine is tested in people?

Before a medicine is tested in people, researchers study whether it could cause serious harm and collect the dose and toxicity information needed to judge whether a human trial is justified. This is preclinical research.

The FDA describes two broad forms of this work: in vitro studies and in vivo studies. The first examines a candidate outside a living organism. The second examines it in a living organism. Neither is a substitute for a human trial. Both are part of the evidence needed before one begins.

Preclinical work is governed by requirements for how studies are run, recorded, and checked. The FDA explains that good laboratory practice requirements cover personnel, facilities, equipment, written protocols, reports, and quality-assurance oversight.

Then comes the Investigational New Drug application, or IND. A sponsor submits preclinical animal-test results and a proposed plan for testing in people. FDA decides whether it is reasonably safe to proceed.

This explains a hard truth about early headlines. A result in a dish or an animal model is a reason to ask the next question. It is not an answer about patients. The 2022 review notes that a candidate can work in cell or animal models yet fail in people because human biology, the true disease target, tissue exposure, dose, and toxicity differ from those models.

What is the difference between Phase 1, 2, 3, and 4 clinical trials?

The phases differ because each tests a different part of the case for a medicine, from early safety to larger evidence of benefit and ongoing safety monitoring after approval. The number of participants grows as the question grows.

The FDA's overview of the drug-review process gives the typical enrollment ranges for the first three phases. Larger trials do not exist to make the process look more serious. They test whether an earlier result holds when more patients, doses, and circumstances enter the picture.

Failure is common at this point. The Acta Pharmaceutica Sinica B review reports that about 9 in 10 candidates entering clinical studies fail in Phase 1, 2, or 3, or during approval. Across 2010 to 2017 clinical-trial data examined in that review, lack of clinical efficacy accounted for 40% to 50% of failures, unmanageable toxicity for 30%, poor drug-like properties for 10% to 15%, and commercial or strategic factors for 10%.

A drug is not judged by its promise. It is judged by its record.

COVID-19 vaccines show why a shorter timeline does not mean a thinner process. NIAID reports that less than three months after SARS-CoV-2 was identified, it began the Phase 1 trial of mRNA-1273 using established trial infrastructure. Years of research on SARS, MERS, related viruses, and vaccine technologies came first, including a four-year NIH-Moderna collaboration on mRNA vaccines for emerging infections.

The safety and efficacy steps remained. Work that often happens one after another happened at the same time: manufacturing began while trials were underway, and existing trial networks combined to enroll thousands of volunteers. Speed came from prior knowledge, funding, infrastructure, and parallel work.

Artificial intelligence changes an earlier part of this path, not the need for the path. A 2024 Drug Discovery Today analysis found 80% to 90% Phase 1 success among AI-discovered molecules in its sample. Its Phase 2 success rate was about 40%, comparable to historical industry averages, and the authors noted the limited sample size. A better starting point still faces the test that matters: results in people.