Strength is essential for carrying out the usual activities of daily life. As there is a loss of strength in elderly, many treatments are based on delaying the loss of strength or maintaining it. Isokinetic dynamometry is the gold standard for assessing strength. It is essential that studies are conducted to allow us to identify the reliability of isokinetic strength assessments in older people. This study aimed to test the absolute and relative intra-session reliability of peak torque and work of a concentric knee extension-flexion performed at 60°/s in elderly. Fifty-two elderly subjects performed three repetitions of bilateral concentric knee extension-flexion at 60°/s using an isokinetic dynamometer. The relative and absolute reliability were calculated between the second and third repetition. The intra-class correlation coefficient values were between 0.94 and 0.98 for peak torque and work in all measures, which is considered “excellent”, except for left leg flexors in women, with values between 0.85 and 0.88, which is considered “good”. The standard error of measurement (SEM) percentage oscillated from 3.9% to 10.5%, with a smallest real difference (SRD) percentage of 10.9% to 29.2% for peak torque. The relative reliability of peak torque and work were excellent for all measures except flexors in women, evidencing the utility of isokinetic dynamometry for monitoring lower limb maximal muscle strength and work of concentric knee extension-flexion at 60°/s/s in the elderly. In addition, an SRD > 19.9% in peak torque and an SRD > 23.1% in work is considered a true change.