The popularity of activity trackers has grown in recent years, and health-conscious consumers are interested in using them for personal motivation and goal setting. Although these devices can estimate several variables, steps are the simplest and easiest to understand. Therefore, the purpose of this study was to investigate the validity of activity tracker step counts during treadmill locomotion (walking and running on a treadmill), overground walking (walking in a hallway), and activities of daily living (ADL). Twenty-one adults (mean ± SD, 26 ± 9 yr) wore six consumer-grade devices (Garmin Vivofit 2, Fitbit Charge, Withings Pulse Ox, Yamax Digi-Walker SW-200, Fitbit Zip, and Omron HJ-322U) and four research-grade devices (two ActiGraph WGT3X-BT [AG] and two StepWatch [SW]). They performed 15 activities for 2 min each, separated by a 1-min rest. Hand-counted (HC) steps served as the criterion measure. Agreement between devices and HC was assessed with 95% equivalence testing. This was done within each category and for all categories combined. For treadmill trials, the SW-modified was significantly equivalent to the criterion, estimating 104.7% of HC steps. For overground walking trials, Fitbit Zip, Yamax Digi-Walker SW-200, and both SW devices were significantly equivalent to the criterion, estimating 98.2% to 102.2% of HS steps. For ADL, the AG-hip (low-frequency extension) was statistically equivalent to the criterion. For all activities combined, the SW-modified (99.5% of HC steps) and AG-hip (low-frequency extension; 94.4% of HC steps) were equivalent to the criterion. In conclusion, a research-grade device (SW) had the greatest accuracy for step counts, making it suitable for clinical research applications. Consumer-grade trackers performed better than expected, and the step count error (10%–15%) may be acceptable for those wanting to track behavior change.