Women's risks for osteoporosis and cardiovascular disease are increasing. Although it is known that these risks decline with physical activity, it is difficult to associate specific risk levels with specific activity levels. Measuring activity levels is complex and problematic, especially in women. The present study explored the validity of a particular measure, the retrospective survey questionnaire, used with a female population to gauge physical activity, both occupational and leisure. To measure construct validity, four retrospective measures were administered to a sample of 43 middle-aged women representing three occupational groups (teachers, clerical workers, nurses) known to differ in activity levels. In addition, measurement approaches were triangulated by including a prospective measure, a beeper-cued record kept by subjects. A significant difference was noted among the groups on retrospective and prospective occupational activity measures, with clerical workers having the lowest levels of energy expenditure. The retrospective occupational measure correlated positively and significantly with the beeper-cued record kept during work; however, no correlation was found between the retrospective leisure measure and the beeper-cued record. These data suggest that the retrospective measure is appropriate for measuring the occupational dimension of physical activity. Additional study is required to identify women's leisure pursuits and the dimension of their household activities.