ObjectiveTo examine the validity of the GT3X® ActiGraph accelerometer and the Google Fit® smartphone application in estimating stepping activity in people with chronic stroke. MethodsThirty-seven stroke survivors walked along a straight, 10 metre hallway over 5 min at their fastest speeds, wearing the GT3X® ActiGraph accelerometer and a smartphone on their paretic lower limb. The criterion-standard measurement made was the actual number of steps, counted by a trained examiner. ResultsThe mean estimated steps measured by the GT3X® ActiGraph and Google Fit® respectively were 276.7 ± 97.6 and 481.0 ± 119.8; that calculated from the examiner's measurements was 472.0 ± 93.9. Statistically significant associations were found between the actual steps and those estimated by the GT3X® ActiGraph (r = 0.56; p < 0.001) and Google Fit® (r = 0.89; p < 0.001). The Google Fit® application demonstrated the highest reliability coefficient (ICC[2,1] = 0.93; p < 0.001; p = 0.37), compared to that of the GT3X® ActiGraph (ICC[2,1] = 0.32; p < 0.001; p < 0.001). ConclusionsThe GT3X® ActiGraph underestimated the data and may not be the most appropriate device to estimate the stepping activity of stroke patients. The findings support the validity of a smartphone application in estimating the stepping activity of individuals with stroke, when worn on the paretic side.