This study aimed to determine the minimum number of repetitions for a high reliability of movement timing in fundamental physical fitness exercises using inertial sensors. Fifteen young men and fifteen women performed eight exercises (two-leg hop, forward lunge, squat, sit-up, shoulder abduction, hip abduction, back extension, and push-up) (preferred tempo, 3 trials, 20 repetitions per trial). The movement timing (cycle of movement in seconds and its phases in seconds and %tcycle) was tested for intra- and inter-trial reliability (SPSS 28.0, p ≤ 0.05). Just two repetitions were adequate for excellent intra- and inter-trial relative reliability (ICCs ≥ 0.75, isolated exceptions only for durations expressed as %tcycle, in only three out of the eight exercises: hip abduction, back extension, and push-up), as well as for high absolute intra- and inter-trial reliability (average SEM% at 5.9%, respectively, and 6.8% and average MDC95% at 13.7% and 15.9%, respectively, which was consistently higher than the upper boundary limit of SEM%, and a rather low CV% ranging from 1.5% to 4.9% and averaging at 3.1%). A total of four repetitions, excluding the initial and the final one, appears adequate for high overall reliability of movement timing in the eight physical fitness exercises examined.