How do bacteria become resistant to antibiotics?

How does resistance spread?

Resistance mechanisms include pumping a drug out, breaking it down, or changing the structure it attacks. When susceptible bacteria are killed or grow poorly, resistant bacteria can multiply. This filtering is natural selection.

A gene is a stretch of DNA carrying instructions. Bacteria pass resistance genes to their descendants. They can also obtain them from other bacteria: horizontal gene transfer. Plasmids, small circles of DNA, can move between cells. A bacterium need not independently invent every defense it gains.

Can we predict a drug-escape mutation?

Sometimes researchers find repeatable routes. In a 2012 laboratory study, separate populations of E. coli evolved resistance to trimethoprim through similar DNA changes in a similar order. With two other antibiotics, the populations reached resistance through more varied combinations of mutations.

A resistance forecast also needs to consider where a gene might come from. If bacteria acquire an existing defense from another cell, a prediction limited to new mutations misses that route. Gene transfer and inheritance therefore matter alongside mutation.

The experiment used one starting strain. Environment and population size can affect available paths; the next mutation in an infection remains uncertain.

Are viruses evolving now?

Yes. A WHO report dated September 25, 2026 described increasing genetic diversity in a circulating influenza group during February–August 2026. That is dated surveillance evidence of continuing change. Viruses and bacteria are different: antibiotics act against bacteria and do not treat viral infections.

Viral forecasting has useful, limited targets. A 2014 influenza model used genetic history and estimates of how well strains spread to forecast the following year’s proportions of related virus groups. A group becoming more common is a different prediction from naming the next new mutation. Such forecasts depend on the information and assumptions used. They estimate possible outcomes, not a guaranteed script.

For the broader question of repeatable evolutionary histories, see our answer on replaying evolution.