As antibiotic resistance spreads, developing sustainable methods to restore the efficacy of existing antibiotics is increasingly important. One widespread method is to combine antibiotics with synergistically acting adjuvants that inhibit resistance mechanisms, allowing drug killing. Here we use co-amoxiclav (a clinically important combination of the -lactam antibiotic amoxicillin and the -lactamase inhibitor clavulanate) to ask whether treatment efficacy and resistance evolution can be decoupled via component dosing modifications. A simple mathematical model predicts that different ratios of these two drug components can produce distinct evolutionary responses irrespective of initial efficacy. We test this hypothesis by selecting Escherichia coli with a plasmid encoded -lactamase (CTX-M-14), against different compositions of amoxicillin and clavulanate. Consistent with our theory, we found that while resistance evolved under all conditions, the component ratio influenced both the rate and mechanism of resistance evolution. Specifically, we found that the current clinical practice of high amoxicillin to clavulanate ratios resulted in the most rapid adaptation to antibiotics via gene dosing responses. Increased plasmid copy number allowed E. coli to increase -lactamase dosing and effectively titrate out low quantities of clavulanate, restoring amoxicillin resistance. In contrast, high clavulanate ratios were more robust - plasmid copy number did not increase, although porin or efflux resistance mechanisms were found, as in all drug ratios. Our results indicate that by changing the ratio of adjuvant to antibiotic we can slow and steer the path of resistance evolution. We therefore suggest using increased adjuvant dosing regimens to slow the rate of resistance evolution.