The aim of this study was to investigate the problem of agonist-antagonist co-contracti during a precision force task performed at different force levels. Using a precision grip, seven young adults performed a constant force matching task (10, 22.5, 35, 47.5, and 60% maximum) as accurately as possible (10 trials per force level). Muscle co-contraction in the thumb was monitored by the surface EMG activity of the extensor pollicis longus (EPL) and the flexor pollicis brevis (FPB), and the ratio between those EMG activities (EPL/FPB). Results showed that both EMG activities increased as grip force increased, but the EPL/FPB ratio decreased over the range of force investigated. Force steadiness (as expressed by the coefficient of variation, CV) appeared as a U-shape function of the force level (with maximal steadiness at 22.5%). Separate analyses at each force level showed no correlation between CV and EMG indices. In addition, the contrast between trials with high and low CV revealed no significant difference in terms of our EMG indices. We conclude that muscle co-contraction and grip force steadiness depend on grip force magnitude, but grip force steadiness does not depend on muscle co-contraction.