If food production is to become more sustainable, weed management systems will become subject to many internal and external constraints; these may sharply limit the available range of weed control methods. If so, it will be imperative to preserve efficacy of acceptable weed control methods by defending these against weed adaptation. Moreover, rapid and ongoing weed adaptation may explain the persistent nature of yield losses to weeds despite technological advances. Research is needed to assess the practical importance of ongoing weed evolution. The first step is to determine the risk of rapid weed adaptation to diverse control measures. Experimentally, this risk can be evaluated by determining heritabilities of traits conferring resistance to control measures, and then using ecological genetic methods to measure the actual magnitude of selection for such traits. Models of multivariate trait evolution can be used to determine the net effect of evolutionary forces bearing on weed adaptation to an integrated weed management system (IWMS). It may also be possible to take a preventive approach by designing agricultural systems so that their large-scale structure impedes weed adaptation to IWMS.