The factorial validity of the Maslach Burnout Inventory (MBI) was studied for 2,931 Canadian teachers (48.2% males and 51.8% females) as a single professional group and for subsamples of this group (1,159 elementary school teachers, 388 intermediate school teachers, and 1,384 secondary school teachers). Study participants were full-time teachers from two large metropolitan areas in central Canada. The subjects represented about 42% of a sample originally sent the study questionnaire. The best-fitting factorial model was cross-validated for samples made by splitting the original group. Exploratory factor analysis and confirmatory factor analysis (CFA) were used to validate the hypothesized three-factor structure. CFA was subsequently used to: (1) identify aberrant scale items; (2) determine, via double cross-validation procedures, the model having the greatest predictive validity; and (3) test for the invariance of measurement and structural relations across calibration and validation samples. Although findings strongly support a three-factor structure, the deletion of four items (numbers 2, 11, 12, and 16) yielded a remarkably improved and psychologically sound instrument. Compared with data for the elementary school and secondary school teachers, model fit for intermediate school teachers was more problematic, although consistent with past research. A 57-item list of references is included, and an appendix gives a breakdown of MBI items on factors of burnout (emotional exhaustion, depersonalization, and personal accomplishment). Four tables present study data. (SLD)