Bergmann's rule predicts a decrease in body size with increasing temperature and has much empirical support. Surprisingly, we know very little about whether 'Bergmann size clines' are due to a genetic response or is a consequence of phenotypic plasticity. Here we use data on body size (mass and tarsus length) from three long-term (1979-2008) study populations of great tits (Parus major), in which there has been a temperature increase, to examine mechanisms behind Bergmann's rule. We show that adult body mass decreased over the study period in all populations and that tarsus length increased in one population. Both body mass and tarsus length were heritable and under weak positive directional selection, predicting an increase, rather than decrease, in body mass. There was no support for micro-evolutionary change and thus the observed declines in body mass were a result of phenotypic plasticity. Interestingly, this plasticity was not in direct response to temperature changes but seemed to be due to changes in prey dynamics. Our results caution against interpreting recent phenotypic body size declines as an adaptive evolutionary response to temperature changes and highlight the importance of considering alternative environmental factors when testing size clines.