This study compared the reliability and validity of different protocols evaluating the force-velocity (F-V) relationship and muscle power in older adults. Thirty-one older men and women (75.8±4.7 years) underwent two F-V tests by collecting the mean and peak force and velocity data exerted against increasing loads until one repetition maximum (1RM) was achieved in the leg press exercise. Two attempts per load were performed, with a third attempt when F-V points deviated from the linear F-V regression equation. Then, the subjects performed 2×3 repetitions at 60% 1RM to compare purely concentric and eccentric-concentric repetitions. The Short Physical Performance Battery was conducted to assess the validity of the different protocols. Significant differences were found in maximal power (Pmax) between mean and peak values and between protocols differing in the number of attempts per load (p<0.01). Registering mean values, a third attempt, and multiple loads (>3), was significantly more reliable (Pmax: CV=2.6%; ICC=0.99) than the other alternatives. Mean values were also observed to be more associated with physical function than peak values (R2=0.34 and 0.15, respectively; p<0.05). No significant differences were observed between concentric and eccentric-concentric repetitions. Thus, collecting mean force and velocity values against multiple loads, while monitoring the linearity of the F-V relationship, seemed to be the more adequate procedure to assess the F-V profile and muscle power in older adults.