A validated physical activity questionnaire for young adults was adapted and validated for use in free living, apparently healthy people, aged 63-80 yr. Test-retest reliability of the questionnaire on 29 participants was 0.89 as determined by Spearman's correlation coefficient. Further classification by tertiles of activity resulted in 72% of the participants being correctly classified and 0% grossly misclassified on two separate occasions. In a similar group of 31 subjects, classifications based on questionnaire activity scores were compared with classifications obtained by repeated 24-h activity recalls and pedometer measurements, showing Spearman's correlations of 0.78 and 0.73, for both methods, respectively. Seventy-one and 67% of the subjects, respectively, were classified in the same activity tertile for both methods. It is concluded that the questionnaire provides a reliable and valid method for classifying elderly subjects into categories of high, medium, and low physical activity.