SCIENCE

95% of Animal-Tested Drugs Flop in People

United StatesThu Nov 19 2026
95% of Animal-Tested Drugs Flop in People

Here is a number worth sitting with: up to 95% of drugs that clear animal testing crash and burn once real people take them. Ninety-five percent. That figure alone should make anyone pause before swallowing the claim that animal experiments are the only road forward. A guest column published Oct. 22 argued exactly that, and the argument holds up poorly.

Biology explains the gap. Dogs, mice, and humans overlap in plenty of ways, but their molecular and metabolic pathways split in ways that matter enormously. Think about something as ordinary as a snack. Chocolate, grapes, and garlic can poison a dog while leaving a person completely fine. If a piece of food can divide species that sharply, why assume a complex drug crosses the same divide cleanly?

Then comes the question of scale. Federal agriculture department records place the number of dogs and cats used in U.S. research facilities at roughly 55,000 to 60,000 every single year. That is not a rounding error, whatever some commentary implies. Every one of those animals carries housing costs, specialized staff, veterinary care, and public funding, all feeding a system with a documented failure rate near 95%. Taxpayers have every right to ask whether that math works.

Non-animal methods, meanwhile, keep getting sharper. New Approach Methodologies, known as NAMs, include organ-on-a-chip systems built to mimic human tissue. Studies indicate these tools already forecast human toxicity and immune responses more accurately than live animal models manage. The technology is still expanding, which is a reason to fund it, not dismiss it. The COMPACT Act, filed as H.R. 10522, would not shut science down overnight. It would move federal money away from low-yield animal models and toward faster, more accurate human-based approaches.

actions