Fifty-two participants walked on a treadmill at 4.8 km/h for 500 steps while wearing four Samsung Galaxy S4 smartphones on the arm, waist, pocket, and hand while each phone simultaneously ran five popular smartphone apps. Actual steps were measured using a hand tally device. Steps were recorded from each smartphone app and compared to the tally counter using repeated measures analysis of variance (ANOVA) tests, and equivalence testing. Of the 20 step measurements recorded (five apps at four locations), all but four (Accupedo at the arm, waist, and pocket; S-Health at the pocket) produced mean underestimations of step counts. ANOVAs showed significant differences between the phone at the hand location for all apps compared to the tally counter (p < .05); three apps had differences at the waist (p < .01), Runtastic had differences at the arm (p < .001), and no differences occurred between the pocket location and the hand tally counter for any of the apps (p > .05). The 90% confidence interval for all apps, except for G-Fit, fell within the equivalence zone for the phone in the pocket while the phone at the hand location included only S-Health within the equivalence zone. Using a Samsung Galaxy S4 smartphone to measure steps at a 4.8 km/h walking pace while carrying the phone in the hand may produce significant errors. However, using the S-Health app while carrying a phone in the pocket appears to provide the most accurate step count in a controlled environment.