Purpose To determine the amount of familiarization sessions required by breast cancer survivors to achieve a reliable measurement of muscle function assessed using isokinetic dynamometry. Methods Twenty-six breast cancer survivors performed three isokinetic knee extension tests separated by, at least, 48 h. The isokinetic testing protocol included one warm-up set of 10 submaximal knee extensions at 120°/s, followed by two sets of four maximal knee extensions at 60°/s, with 2-min rest interval between sets. Peak torque (PT), time to peak torque (TPT), angle of peak torque (APT), and average power (AP) of each trial was used for the assessment of testing reliability. Percentage change in the mean, typical error, coefficient of variation and intraclass correlation coefficients (ICC 2.1 ) were calculated to determine test-retest reliability. Results For PT, change in mean was lower between trials 2 and 3 than between trials 1 and 2 (4.18% and 13.18%, respectively), and ICC was greater between trials 2 and 3 than between trials 1 and 2 (0.962 and 0.818, respectively). For TPT and APT, ICC was clinically acceptable only between trials 2 and 3 (0.757 and 0.803, respectively). For AP, change in mean was clinically acceptable between trials 2 and 3 (9.84%), while ICC met acceptable reliability between both, trials 1 and 2 and, trials 2 and 3 (0.756 and 0.891, respectively). Conclusion At least one familiarization session is adequate to achieve reliable measurements of muscle function using isokinetic dynamometry, while avoiding the impact of learning effect of the measurements in breast cancer survivors.