New drugs are discovered through a multi-step process that typically takes 10 to 15 years and costs between $1 billion and $2 billion per successful drug. That cost includes all the candidates that fail along the way. The journey begins with identifying a biological target in the body, moves through rounds of lab testing and animal studies, and culminates in human clinical trials before a regulatory agency like the FDA decides whether to approve it.
Finding the Right Target
Every drug starts with a question: what’s going wrong in the body, and where can we intervene? Scientists look for a specific protein, enzyme, or receptor that plays a key role in a disease. They find these targets through two main routes. One is genetic association studies, where researchers compare the DNA of people with a disease to people without it, looking for genes that show up more often in the sick group. The other is phenotypic screening, where researchers expose cells or tissues to various conditions and observe what changes, working backward to figure out which molecule is responsible.
Once a potential target is identified, it needs to be validated. Scientists examine the protein’s three-dimensional structure using X-ray crystallography to determine whether a small molecule could physically bind to it and alter its behavior. They also develop lab assays (essentially standardized tests) that can measure whether a compound is actually hitting that target. And they look at genetic data from patients to predict whether blocking or activating the target might cause unwanted side effects. A target that passes all three checks moves forward.
Screening for Compounds That Work
With a validated target in hand, researchers need to find a chemical compound that interacts with it in the right way. This is where high-throughput screening comes in. Automated systems test thousands of compounds against the target, flagging any that produce a measurable effect. A compound counts as “active” if its signal stands out significantly from the baseline noise of the experiment.
Those initial hits are then retested at multiple doses to confirm they’re real. Researchers also filter out compounds that are toxic to cells, that react nonselectively with many targets instead of just the intended one, or that work through crude chemical mechanisms rather than precise binding. The survivors get grouped by their chemical structure, and scientists look for patterns: what structural features make a compound more potent, more selective, or more likely to behave well in the body? This analysis narrows the field to a handful of “lead” compounds worth investing real chemistry resources into optimizing.
Optimizing a Lead Compound
A lead compound rarely becomes a drug as-is. Chemists systematically modify its structure, tweaking atoms and functional groups to improve potency, reduce toxicity, and make it more likely to survive the journey through the human digestive system and bloodstream. They’re balancing competing demands: the molecule needs to dissolve well enough to be absorbed, remain stable long enough to reach its target, bind tightly enough to have an effect, and then be cleared from the body without causing harm.
This phase can take years of iterative cycles. Each new version gets tested in lab assays and compared to the previous one. The goal is a compound that’s effective at low doses, selective for its target, and has properties that make it viable as a pill, injection, or other deliverable form.
Preclinical Testing
Before any compound can be tested in humans, it must go through preclinical safety studies. These typically involve both cell-based experiments and animal studies designed to reveal how the drug behaves in a living system: how it’s absorbed, where it accumulates, how it’s broken down, and what toxic effects it might cause.
Safety programs normally include testing in two relevant animal species, though in some cases one species is sufficient if it’s the only one whose biology is close enough to humans for the results to be meaningful. Researchers evaluate effects on critical organ systems, including cardiovascular and respiratory function. The specific tests depend on the type of drug. For biologically derived drugs like antibodies and proteins, standard genetic toxicity tests and long-term cancer studies aren’t typically required because those tests were designed for traditional chemical compounds and don’t translate well. The FDA takes a case-by-case approach, expecting the testing program to match the science of each individual drug.
Clinical Trials in Humans
If preclinical results look promising, the drug developer files an application with the FDA to begin human testing. Clinical trials unfold in three main phases, each with a different purpose and scale.
Phase I trials enroll 20 to 80 people, usually healthy volunteers. The primary goal is safety: researchers start with very low doses and gradually increase them, watching for side effects and learning how the body processes the drug. Phase II trials expand to 100 to 300 people who actually have the disease the drug is meant to treat. Here, researchers are looking for the first real evidence that the drug works, while continuing to monitor safety. Phase III trials are the large confirmatory studies, enrolling 1,000 to 3,000 patients. These compare the new drug to existing treatments or a placebo, generating the statistical evidence regulators need to decide whether the drug’s benefits outweigh its risks.
Most drugs that enter clinical trials never make it through all three phases. They fail because they turn out to be less effective than expected, produce unacceptable side effects, or don’t perform better than treatments already on the market. This high failure rate is a major reason drug development is so expensive: the cost of every failure gets folded into the price of every success.
Regulatory Review and Approval
After successful Phase III trials, the drug developer submits a New Drug Application to the FDA. This document tells the drug’s entire story: the results of animal studies, clinical trial data, the drug’s ingredients, how it behaves in the body, and how it’s manufactured, processed, and packaged. FDA reviewers comb through all of this to determine whether the drug is safe and effective for its intended use.
The bar for approval isn’t perfection. It’s whether the drug’s benefits justify its risks for the specific population it’s designed to treat. In 2025, the FDA approved 46 novel drugs, meaning drugs with active ingredients never before marketed in the United States.
How AI Is Changing the Process
Artificial intelligence is reshaping several stages of drug discovery. Machine learning models can analyze massive biological datasets to identify potential drug targets faster than traditional methods. Natural language processing tools scan millions of scientific papers and patent filings to find connections between genes, proteins, and diseases that human researchers might miss.
AI also makes high-throughput screening more efficient. Instead of physically testing every compound in a library, machine learning models can predict which compounds are most likely to hit a target, allowing researchers to prioritize their screening efforts. Once hits are identified, AI helps refine candidates by predicting how structural changes will affect a compound’s potency, stability, and safety profile. The technology doesn’t replace the experimental work, but it compresses timelines by reducing the number of dead ends researchers have to explore.
Drug Repurposing: A Faster Path
Not every new drug starts from scratch. Drug repurposing takes medications already approved for one condition and investigates whether they work for something entirely different. This approach is faster and cheaper because the safety profile of the drug is already well established. Repurposed drugs can skip much of the preclinical development phase and move directly into clinical trials for the new use.
The most common method is to screen libraries of approved drugs against a new target. In one study, researchers screened a library of 853 FDA-approved drugs to find ones that could kill a specific type of antibiotic-resistant bacteria. They used an assay that detects a molecule released when bacterial cells die, allowing them to rapidly identify which existing drugs had unexpected antibacterial effects. This kind of systematic screening has become a standard tool for finding new uses for old drugs, and it’s particularly valuable for diseases where traditional drug development is too slow or too expensive to attract investment.