A representative sample of 420 subjects aged 55 to 84 was sought for participation in a study of the relationship between physical activity and health. A stratified random sample with 35 subjects in each of 12 age-gender categories was designed, using lists generated consecutively until the target number for each category was reached. Although the protocol required walking and treadmill tests, 60.3 percent of those eligible participated, a figure comparable to that of NHANES II and superior to that of the Canada Fitness Survey. Participants were younger, more likely to be male, less likely to be currently married, and more likely to have had a white-collar job and some postsecondary education than were nonparticipants. Difficulties of obtaining subjects and methods for avoiding these difficulties include offering many opportunities to attend and determining accurately the reason for nonparticipation.