You must always finish the full course of antibiotics, even once you feel better — stopping early is what creates antibiotic resistance.
Where the rule came from
The instruction is old, and its origin is more rhetorical than experimental. It traces back to the earliest era of antibiotic use, when the fear was treatment failure — an infection knocked back but not cleared, relapsing worse than before.1
Somewhere along the way the justification quietly swapped. The advice stayed identical, but the reason given became resistance — and that reason does not follow from how resistance actually arises.
What the evidence actually shows
Resistance is a selection process. Antibiotic exposure kills susceptible organisms and leaves resistant ones with less competition. More exposure means more selection pressure, not less.2
Over the past two decades a substantial body of randomised trials has compared shorter courses with the traditional longer ones. The pattern is consistent: for many common infections, shorter courses are non-inferior on cure rates, with fewer adverse effects.3
Where the old rule still holds
Duration matters enormously in a specific set of infections — tuberculosis above all, and also endocarditis, osteomyelitis and prosthetic-material infections. In these, stopping early is genuinely dangerous.
What makes this myth so durable is that it sounds like caution. Finishing the pack feels responsible. But every extra day of exposure is another day of selection pressure on the trillions of bacteria that were never the target. The rule isn't wrong so much as it's answering a question nobody asked: it protects against relapse, not resistance.
Verdict
Outdated — the reason given is wrongCompleting a prescribed course is still the right default, but "it prevents resistance" is not why. Duration is infection-specific, several standard courses have been shortened on trial evidence, and a handful of infections remain absolutely duration-critical.
Sources
- Historical review of antibiotic course-length guidance — [Journal, Year]
- Selection pressure and duration of exposure — [Journal, Year]
- Short vs. standard course non-inferiority trials — [Journal, Year]
No corrections to date. If you can show me I am wrong, with a source, it gets fixed here and dated — send it over.