This study compares the accuracy of a two-dimensional accelerometer worn on the ankle (a step activity monitor) with that of an electronic, digital pedometer worn on the belt line. Twenty-nine human subjects were evaluated while they briskly walked 400 M, slowly walked 10 M, and ascended and descended a flight of stairs. The step activity monitor had less error in all activities; its mean absolute error was 0.54%, whereas that of the pedometer was 2.82%. The difference was more pronounced in obese subjects (body mass index greater than 30), with an overall mean absolute error of 0.48% for the step activity monitor and 6.12% for the pedometer (nearly 13 times that of the step activity monitor). For subjects with a body mass index less than 30, the step activity monitor had an overall error of 0.56% and the pedometer had an overall error of 1.56% (less than 3 times that of the step activity monitor). The absolute error of the pedometer was positively correlated with body mass index (r = 0.792, p < 0.0001) and weight (r = 0.753, p < 0.0001), whereas the error of the step activity monitor was not significantly correlated with either. Neither device was significantly biased by age, gender, or the presence of a lower-extremity joint prosthesis. The accuracy and additional capabilities, including a real-time memory record of activity, of the step activity monitor make it well suited for objectively quantifying ambulatory activity, especially for obese subjects.