After discussing various problems with the chivalry hypothesis and labeling theory -- the two major explanations of gender disparities -- an empirical test is made to determine the accuracy of both perspectives. An analysis of 543 adult felony cases in Dade County, Florida, using multiple regression to control for the effects of relevant legal and nonlegal variables, indicated inconsistent differences in treatment by sex at the levels of negotiations, prosecution, conviction, and sentence. These differences also vary over the three time periods examined: 1965-1966, 1971, and 1975-1976. The findings do not support any existing theory of the differential handling of male and female offenders, clearly indicating a need for the development of theory that takes into account the growing body of empirical work in this area. (Publisher abstract)