Concentrated chemical spills have been shown to impact adversely on fish populations and even cause localized population extinctions. Evaluating population-level impacts of sublethal exposure concentrations is, however, complex and confounded by other environmental pressures. Applying effect measures derived from laboratory-based chemical exposures to impacts in wild fish populations is constrained by uncertainty on how biochemical response measures (biomarkers) translate into health outcomes, lack of available data for chronic exposures and the many uncertainties in available fish population models. Furthermore, wild fish show phenotypic plasticity and local adaptations can occur that adds geographic and temporal variance on responses. Such population-level factors are rarely considered in the chemical risk assessment process and can probably be derived only from studies on wild fish. Molecular technologies, including microsatellite and SNP genotyping, and RNASeq for gene expression studies, are advancing our understanding of mechanisms of eco-toxicological response, tolerance, adaptation and selection in wild populations. We examine critically the application of such approaches with examples including using microsatellites that has identified roach (Rutilus rutilus) populations living in rivers contaminated with sewage effluents that are self-sustaining, and studies of stickleback (Gasterosteus aculeatus) and killifish (Fundulus heteroclitus) that have identified genomic regions under selection putatively related to pollution tolerance. Integrating data on biological effects between laboratory-based studies and wild populations, and building understanding on adaptive responses to sublethal exposure are some of the priority research areas for more effective evaluation of population risks and resilience to contaminant exposure.