Performance on the Rapid Alternating Stair Tap Test (RASTT) relates well with lab-based measures of medial-lateral (ML) motion and consistency of step timing in older adults. It may have clinical utility for fall risk assessment. The RASTT forces ML weight-shifting motion due to the need to alternate single-leg stance from side-to-side while alternating foot tapping on a step. The duration of the test is of interest; the test aims to evaluate rapid movement capability and needs to be short enough to allow rapid performance to be maintained. PURPOSE: To determine whether stepping characteristics change during the 20-second RASTT. METHODS: Older adult participants (n = 75, mean age: 79.8 years, age range: 70-96 years) with a broad range of functional capacity participated in this study. Acceleration was measured with a triaxial wireless accelerometer affixed to a belt around the lower back while participants alternated each foot tapping the center of a step in front of them and then stepping back to the floor as many times as safely possible within 20 seconds. Three familiarization trials were conducted, with one minute of rest in between. Signal events were used to identify steps and calculate mean step time during the trial, the standard deviation (SD) of step time, and acceleration magnitude and variability (mean and SD of the per-step peak-to-peak ML acceleration). The difference in these outcome measures between the first and second half of the 20-second testing period was evaluated using a paired samples Wilcoxon test, as the data were not normally distributed. RESULTS: Mean step time and mean per-step peak-to-peak ML acceleration did not differ between the first and the second half of the trial. Small reductions occurred between the first to the second half of the trial for step time SD (median [IQR]: 0.061 [0.071] to 0.047 [0.064] s, Z = -2.054, p = 0.04) and SD of the per-step ML peak-to-peak acceleration (0.143 [0.115] to 0.142 [0.145] g, Z = 2.017, p = 0.044). CONCLUSION: Rhythmicity of stepping increased and the ML acceleration variability reduced over the course of the 20-second trial. This indicates that the duration of the trial allowed the control of motion to stabilize over the course of the test, observed as less variable step time and less variable peak ML acceleration, without being long enough to cause step time to slow.