The aim of this study was to summarize both practical and methodological issues in using pedometers to assess physical activity in a large epidemiologic study. As part of a population-based survey of cardiovascular disease risk factors, physical activity was assessed using pedometers and activity diaries in 775 men and women ages 25-64 years who were residents of Burnie, Tasmania, 1998-99. Common data problems were classified by type. The frequency of each problem and the methods used to identify it are reported along with strategies to correct or prevent each problem type. Pedometer data from 15 (1.9%) participants could not be used due to errors in completing the pedometer protocol. Among 760 participants with usable data, the median number of steps was 9,729 for men and 10,388 for women. Pedometer steps per day were modestly correlated (r = .20, p < .0001) with the duration of pedometer wear, which ranged from 4.50 to 21.75 hr. Adjustment for wear time, however, did not alter observed correlations between pedometer steps and cardiovascular risk factors. The authors conclude that pedometers can be used in large population studies with a relatively low frequency of data errors. However guidelines for consistent data collection and interpretation are needed.