[BDSP. Notice produite par INIST-CNRS DTR0xA44. Diffusion soumise à autorisation]. Few studies have evaluated the reliability of a self-administered current and historical physical activity questionnaire (PAQ) among middle-aged and elderly women. We evaluated the reliability of a self-administered PAQ designed to assess total (24 h) current and historical physical activity at age 15,30, and 50 years, which was completed by a subgroup of 303 women aged 56-75 years from the Swedish Mammography Cohort (SMC). Total physical activity covered occupational and household activity as well as walking/bicycling, exercise, and inactivity (sitting watching TV/reading). Reliabilities (1-year test-retest) of continuous activity measures in metabolic equivalents were evaluated using an intraclass correlation coefficient (ICC) ; classification consistency was evaluated using sensitivity and specificity statistics. The ICC for total current physical activity was 0.69 ; for specific types of activities the ICC ranged from 0.49 to 0.59 ; for historical physical activity, the reliabilities for total activity ranged from 0.75 for age 50 to 0.81 for age 30 years, being substantial for occupational activities (ICC=0.73-0.75), and fair to moderate for inactivity (ICC=0.31-0.60). These data indicate that the PAQ used in the SMC is a satisfactory and reproducible measure of current and historical physical activity, for total as well as for different types of activities.