A number of studies suggest that the gender of a legislator affects his or her congressional ideology. We argue that these studies may have produced misleading results because of insufficient controls for constituency influences. To better account for constituency effects, we use a longitudinal research design based on electoral turnover, which holds constituency constant while allowing gender and party to vary. We apply ordinary least squares regression to data from the 103d, 104th, and 105th Houses of Representatives and estimate the effect of gender turnover on changes in DW-NOMINATE roll-call voting scores. We find that, when we sufficiently control for both party and constituency influences, gender is not a determinant of the liberalness of a representative's roll-call voting behavior.