SigmaPCB and p,p'-DDE levels within and among walleye (Stizostedion vitreum) populations were examined to determine how the method of data analysis could influence the interpretation of (i) gender differences and (ii) geographic variation. In the lower Great Lakes (Huron, Erie, and Ontario) whole-body burdens of both contaminants tended to increase with body mass at a faster rate in males than in females. Thus, males generally had higher burdens than females at large body sizes but not at small body sizes. This result was not strongly influenced by the method of expressing contaminant level (burden, wet mass concentration, or lipid mass concentration) but was influenced by the choice of covariate (body mass, body length, or age) in some cases. Mean sigmaPCB and p,p'-DDE concentrations of walleye muscle declined along a gradient from the lower Great Lakes to the Northwest Territories. Analyses using means adjusted for age yielded a stronger contrast between Great Lakes and non-Great Lakes populations than analyses using means adjusted for body length. The gender composition of fish samples and the type and level of covariate used in statistical analyses should be considered in studies of spatiotemporal variation in organochlorine bioaccumulation in fish.