Selective harvesting, like fishing, can erode size diversity in wild fish populations by truncating population size-structures and homogenising growth rates. Many important population and fishery metrics are influenced by size, including reproduction, recruitment success, stock biomass and mortality regimes. Reductions in trait variation (within the population portfolio) can lead to altered population dynamics, with stocks becoming more synchronous within a region and hence more susceptible to environmental perturbations. Understanding the capacity to which exploited populations can buffer environmental disturbances through portfolio effects is becoming increasingly important due to the impacts of climate change. To test whether fishing and climate warming have eroded trait variation in commercially fished stocks, we analysed time series (up to 55 years) of size-at-age data for 20 species from two large marine ecosystems (LMEs), in relation to life stage (juvenile or adult), fishing pressure and temperature. We measured three different metrics to quantify changes in size-at-age: the mean, maximum and variance in size-at-age. We found clear evidence that size-at-age changed through time in both LMEs and that these temporal changes could be attributed to both temperature and fishing. The magnitude of change was dependent on life stage, with results suggesting that warming waters can result in a reduction in adult body size but that this effect is potentially mitigated by a fishing-induced relaxation of density-dependent controls. Our results provide novel insight into the importance of addressing the synergistic impacts of fishing and climate on body size. Such knowledge will improve predictions of population vulnerability to disturbance leading to more sustainably managed fisheries, especially in the face of rapid environmental change.