Gottfredson's (1981) developmental theory of career choices was evaluated using longitudinal data. This study builds on previous research into Gottfredson's model by addressing methodological issues in the confounding of sex type, prestige, and interest measures when measuring expressed career choices by statistically controlling for the inter-relations among these variables in the analyses. Participants were 2,353 women enrolled at the Seven College Conference Schools and 321 men enrolled at Harvard and Vassar. Expressed career preferences were coded for sex type, prestige, and the Things/People and Data/Ideas interest dimension. Stability of expressed career choices over a one year period and discrepancies between current career choices and idealized aspirations were analyzed using partial correlation analyses to control for the non-independence of the constructs. Gottfredson's (1981) theory was not supported for either men or women, because prestige and the Things/People dimension of interest were found to be more stable than sex type. Implications for re-conceptualizing Gottfredson's theory are discussed, along with implications for research and practice.