Density-independent redistribution in response to changing temperature conditions and density-dependent redistribution in response to changing temperature preferences are two contrasting hypotheses to explain changes in the distribution of cod populations. I tested these hypotheses using survey data on the distribution of cod during the feeding season in the southern Gulf of St Lawrence. These tests used indices of cod distribution as the dependent variable and indices of cod abundance and bottom temperature as explanatory variables. Both hypotheses were often supported in univariate tests that ignored confounding between the two explanatory variables. Tests that accounted for this confounding indicated an effect of cod abundance on distribution for ages 4–8+ years, and 3+, but provided no support for an effect of environmental conditions on distribution except for age 3. Variation in the temperature distribution of cod was consistent with an effect of abundance on interannual variation in distribution and with no effect of environment except for age 3 cod. Age 3 cod were consistently under-represented in areas of very cold bottom water but older cod were not. As predicted by bioenergetic considerations, cod tended to occupy colder water at high levels of abundance, suggesting a mechanism for density-dependent shifts in distribution. Shifts in cod distribution during the feeding season in the southern Gulf appear to be more closely linked to density-dependent changes in environmental preferences than to density-independent responses to changing environmental conditions.