Numerous studies have examined the validity of activity monitors for the assessment of physical activity in different populations. Less work has been done to establish the reliability of these devices for use in field-based research. Test-retest values have been reported for two units worn simultaneously but generalizations to all monitors may not be appropriate if only two monitors are evaluated. The purpose of this study was to systematically examine the intra-unit and intra-individual variability of activity monitors for a structured bout of physical activity. Participants were 32 (24 female and 8 male) college students. Each participant completed three trials of treadmill walking at 3 mph while wearing 10 different CSA activity monitors worn on the right hip. For each trial, the participant completed 10, 5-minute bouts of walking, with 1-minute of standing rest between each trial. The monitors were initialized to record data in 60-second epochs and each trial was started at the beginning of a new minute to ensure proper synchronization of the data. Coefficients of variation (CV=SD/Mean) were used to examine the variability in the scores across monitors and trials. Using data from all 96 trials, the mean CV across the 10 monitors was 8.90% +/- 4.6%. The mean CV for each individual’s set of three trials was similar (10.4% +/- 1.2%). This variability was consistent for the 10 monitors with mean CV values ranging from 8.7% to 12.3%). The fact that variability was just as high when examined separately for each individual suggests that much of the variability in output is due to differences across trials rather than to specific differences between the monitors. Greater variability was found when comparisons were made across individuals for a given monitor (mean CV=20.0%). This suggests that individual characteristics such as height and weight account for considerable variability in the output even when participants perform the same task. The single measure intraclass reliability coefficient (computed using a one-way ANOVA procedure) ranged from R=.74 to .81 for the three trials but the average measure coefficients were considerably higher (R=.97 to R=.98). Because field based data is obtained from one trial and one monitor, the single measure coefficients reflect the expected reliability of the monitors. This variability may have implications for the design of studies using activity monitors to assess free-living physical activity.Keyword(s): assessment, physical activity, research