The reliability and validity of three physical activity questionnaires were studied using 90 Flemish males (30 aged 30 years, 30 aged 35 years, and 30 aged 40 years). Intraclass correlations ( R ) and kappa values were calculated to verify within judge and between judges reliability (objectivity) and test-retest reliability (stability) of the Tecumseh Community Health Study Questionnaire, the Five City Project Questionnaire, and the Baecke Questionnaire. Results showed high for within judge and between judges reliability. R coefficients for stability varied between 0.47 and 0.95. Kappa values varied between 0.20 and 0.73. Concurrent validity was investigated by comparing three levels of professional status. Workmen had higher indices concerning physical activity during work than clerks and managers. Congruent validity, studied by means of principal-components analysis, confirmed subdivision of habitual physical activity into three entities, physical activity during work, sports activities, and general leisure time. These results indicate that reliable and valid data can be obtained in Flemish males by three interviewer-assisted physical activity questionnaires. Am J Epidemiol 1998; 147: 982–90.