Abstract Background The prevalence of antibiotic resistance continues to rise, rendering many valuable antimicrobial drugs ineffective. Pairwise cyclic antibiotic therapy, where treatment is rapidly switched between two antibiotics, has been demonstrated in vitro to limit the evolution of antibiotic resistance. However, what happens when resistance inevitably evolves to one of the drugs? Methods In this study, we perform over 450 evolution experiments to test the resilience of four proposed cyclic therapies. We use soft agar gradient evolution and ‘flat plates’ to identify resistance trade-offs that are resilient to compensatory mitigation. Resensitizations were detected by antimicrobial susceptibility assays, and their mechanistic underpinnings were elucidated via genomic and phenotypic analyses. Results Resistance evolves readily and collateral sensitivity (CS) (where resistance to drug A leads to hypersensitivity to drug B) does not hinder the evolution of multidrug resistance and does not predict or promote resensitization. However, if resistance to drug B increases susceptibility to A, a phenomenon we term backward CS, resistance to A can be reduced or even reversed. For example, we show that Escherichia coli cells frequently become hypersensitive to β-lactams upon aminoglycoside resistance acquisition, due to conflicting modifications to the proton motive force and efflux pumps. We also find for the first time that polymyxin B resistance can be entirely reversed by exposure to tigecycline, through the acquisition of compensatory mutations that reduce the fitness penalty of tigecycline resistance. Conclusions The longevity of drug cycling protocols can be significantly improved by leveraging backwards CS to resensitize cells as antibiotic resistance evolves.