The purpose of this study was to assess the repeatability of a dual gradient-recalled echo (GRE) muscle functional MRI technique. On two days, subjects (n=8) performed 10 s isometric dorsiflexion contractions under conditions of: 1) maximal voluntary contraction (MVC), 2) 50% MVC (50%MVC) or 3) 50% MVC with concurrent proximal arterial cuff occlusion (50%MVCcuff). Functional MRI data were acquired using single-slice dual GRE (TR/TE=1000/6, 46 ms) EPI for 20 seconds before, during, and for 180 seconds after each contraction. The mean signal intensity (SI) time courses at each TE (SI6 and SI46, reflecting variations in blood volume and %HbO2, respectively) from the tibialis anterior (TA) and extensor digitorum (EDL) muscles were characterized with the post-contraction change in SI and the time to peak SI (ΔSI and TTP, respectively). ΔSI6 and ΔSI46 were 36% and 31% higher following an MVC than after a 50%MVC (p = 0.05 and p = 0.07 respectively). For ΔSI6 the highest intraclass correlation coefficients (ICC) were observed for the TA muscle at the 50%MVC and MVC condition, with values of 0.83 (p = 0.01) and 0.88 (p = 0.005), respectively. Bland-Altman plots revealed repeatability coefficients (RC) for the 50%MVC and MVC conditions in the TA muscle of 1.9 and 1.4, respectively. The most repeatable measures for ΔSI46 were obtained for the 50%MVC and MVC conditions in the EDL muscle (p = 0.01 and p = 0.04, respectively). Bland-Altman plots revealed RC’s for 50%MVC and MVC conditions in the EDL muscle of 3.9 and 5.7, respectively. ΔSI6 and ΔSI46 increased as a function of contraction intensity. The repeatability of the method depends on the muscle and contraction condition being evaluated, and in general, is higher following an MVC.