Purpose: The purpose of this study was to determine test–retest reliability of Global Positioning System (GPS) units for measuring distance traveled when walking and cycling (AT), and to determine whether GPS unit placement influences accuracy. Methods: Participants (N = 19) completed two walking and cycling trials at self-selected speeds on a measured 1489-m course wearing two Garmin GPS units, worn in lanyard and waistband placements. GPS estimates of travel distance were compared with actual distance, and test–retest reliability was examined. Data-cleaning protocols were developed to remove signal noise. Results are presented for both raw and cleaned data. Results: For both raw and cleaned data, no significant differences were observed between trials (trial 1 vs trial 2), unit placement (lanyard vs waistband), or AT mode (walk vscycle) (P Q 0.05). Both lanyard and waistband units significantly overestimated distance traveled during walking trials (P e 0.05), but not cycling trials (P Q 0.05). The relative technical error of measurement (TEM) of the raw data ranged from 3.74 to 15.51%, and average absolute errors ranged from 5.03 to 8.53% for all trials. A significant position by AT mode interaction was observed for clean data (P G 0.05). Relative TEM for the clean data ranged from 1.42 to 1.98%, and average absolute errors ranged from 0.32 to 1.97%. Intraclass correlations (ICC) were poor to fair for all trials using raw and cleaned data. Conclusion: Signal noise during unitinitialization may adversely affect unit performance; however, application of data-cleaning procedures to remove data associated with signal noise improves unit ability to measure distance. Results suggest that the lanyard position is the optimal placement for units during data collection.