The threat of drug-resistant bacteria grows more pressing with every year. These microbes can shrug off the most potent antibiotics, including some drugs of last resort. Some bacteria have become resistant to all of our available drugs. Scary stuff, but bacteria don’t have to resist antibiotics to defy them. There is another way—a much simpler, very common, and largely unappreciated one.

The vast majority of antibiotics work by killing bacteria that are actively growing and multiplying. Think of the antibiotics as wrenches thrown into the midst of whirring machines. If the machines are off—their cogs still, their motors silent—the wrenches have no effect. So it is with bacteria. By simply doing nothing, and entering a dormant or extremely slow-growing state, they can survive. They can keep their heads down until the antibiotic has diffused away and the danger has passed.

This strategy is called tolerance, and the cells that practice it are called persisters. It’s very different from resistance. Resistant microbes have special genetic tools that allow them to flout antibiotics: They build pumps that expel the drugs, or enzymes that destroy them. Persisters have no such innate tricks; if they start growing again in the presence of an antibiotic, they’d die. It’s their behavior that saves them—and that behavior is common to many microbes. Hit a colony with antibiotics and most will die, but a small fraction will persist. When the drugs disappear, they can rebound, which is partly why many infections are so hard to treat.

Tolerance and resistance may be different strategies, but they are also connected. As Irit Levin Reisman, from the Hebrew University of Jerusalem, has now demonstrated, the former can lead to the latter.

Reisman exposed the common gut microbe E. coli to the antibiotic ampicillin, at doses commonly used to treat people. After several cycles of exposure, as expected, the bacteria evolved to be resistant; they developed mutations in a gene called ampC, which allowed them to destroy ampicillin. By the experiment’s end, the team needed a million times more of the drug to restrain the bacteria.

By tracing the history of these microbes, the team noticed that the resistant strains only ever emerged from tolerant ones. First, the microbes developed mutations that slowed their growth, allowing them to persist through the waves of ampicillin. Only then did they acquire the crucial ampC mutations that they used to defuse the drug itself. “It was very clear,” says Nathalie Balaban, who led the study. “Each time we got resistance, the bacteria had become tolerant before.”

“That’s an unanticipated link,” says Kim Lewis, from Northeastern University. “Many people, including me, have been arguing that tolerance allows the pathogen to survive and live to fight another day. And if you have a surviving population, sooner or later, resistance will develop. But what [Balaban and her team] have shown is that when the resistance mutation occurs in the background of tolerance, the pathogen survives better.”

After all, why don’t the bacteria just evolve resistance straight away? Why do they always try tolerance first? Balaban has two answers, which she confirmed through simulations.

First, resistance is harder. To become resistant, bacteria typically need to change particular genes like ampC in specific ways. To become tolerant, they just need to slow their growth, and there are many genes they could change to do so. There are many paths to tolerance, but only a few to resistance.

Second, resistance isn’t absolute. It’s very rare for a bacterium to become completely invincible because of a single mutation; more likely, it just becomes a little tougher. “Resistance mutations occur all the time, but they disappear from the population because their owners still have some probability of dying,” explains Balaban. “But if they those bacteria are also tolerant, there’s a much higher probability that their mutations won’t be lost.”

Scientists need to be aware of that, says Tami Lieberman from MIT, because many of them are scanning the genomes of infectious bacteria to build a catalog of resistance mutations. But if resistance follows tolerance, those scans would identify tolerance mutations as well. “We need to expect this,” says Lieberman, so that those mutations aren’t discarded as being unimportant. They clearly are, and they pave the way for classical resistance.

“The million-dollar question is whether this occurs in the clinic, in patients on antibiotics,” says Neeraj Dhar, from the Swiss Federal Institute of Technology in Lausanne. “If it was, one would expect that the resistant strains in circulation would also carry mutations in genes implicated in persistence. This is easily verifiable, and it would suggest that targeting persistence would be a way of combatting or at least delaying the development of resistance.”

Many scientists have already been developing methods of eradicating persisters. In 2011, James Collins identified a way of persuading these sleeper cells to take up drugs that then kill them. Two years later, Lewis’s team discovered a chemical that kills persisters by forcing them to eat themselves in their sleep. It cleared severe infections in mice, and a company called Arietis Pharma, which Lewis founded, is now optimizing the drug so it can be tested in human trials.

The bigger problem is that “there is no test for tolerant bacteria,” Balaban says. “There are already some antibiotics that are active against tolerant bacteria, but they usually have side effects, so you really need to use them in cases where tolerance is a problem.” And her team is now working on tests that can identify such cases.