Based on three independent samples of high school adolescents in central Canada, aged 12 to 18 years, exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were used to identify, test, and cross-validate first-order and second-order factorial structures underlying the Beck Depression Inventory (BDI). Specifically, the study focused on: (1) determining an initial first-order structure of the BDI based on Group 1 (n=196); (2) postulating and testing alternative first-order structures using CFA procedures based on Group 2 (n=284), on the basis of EFA findings; (3) testing for the validity of a hierarchical structure that includes a single second-order factor of general depression; and (4) cross-validating plausible first-order and second-order structures across Group 3 (n=205) using double cross-validation procedures to identify models having the highest degree of predictive validity. Based on these three increasingly stringent tests of the BDI, the findings suggest that, for non-clinical adolescents, the most appropriate factorial structure comprises a higher-order general factor of depression, with three first-order factors--negative attitudes, performance, and somatic elements. Four data tables are included. (Author/TJH)