In the context of global changes, fish are increasingly exposed to multiple stressors that have cascading effects from molecules to the whole individual, thereby affecting wild fish populations through selective processes. In this review, we synthetize recent advances in molecular biology and evolutionary biology to outline some potentially important effects of stressors on fish across biological levels. Given the burgeoning literature, we highlight four promising avenues of research. First, (1) the exposure to multiple stressors can lead to unexpected synergistic or antagonistic effects, which should be better taken into account to improve our predictions of the effects of actual and future human activities on aquatic organisms. Second, (2) we argue that such interactive effects might be due to switches in energy metabolism leading to threshold effects. Under multiple stress exposure, fish could switch from a "compensation" strategy, i.e. a reallocation of energy to defenses and repair to a "conservation" strategy, i.e. blocking of stress responses leading to strong deleterious effects and high mortality. Third, (3) this could have cascading effects on fish survival and population persistence but multiscale studies are still rare. We propose emerging tools merging different levels of biological organization to better predict population resilience under multiple stressors. Fourth (4), there are strong variations in sensitivity among populations, which might arise from transgenerational effects of stressors through plastic, genetic, and epigenetic mechanisms. This can lead to local adaptation or maladaptation, with strong impacts on the evolutionary trajectories of wild fish populations. With this review, we hope to encourage future research to bridge the gap between molecular ecology, ecotoxicology and evolutionary biology to better understand the evolution of responses of fishes to current and future multiple stressors in the context of global changes.