Force estimation based on electromyography (EMG) has been proven to be useful for deriving proportional control for myoelectric devices. Muscle synergies seem to be relevant for force estimation since they are patterns of co- activation of muscles during actions. This study investigates the use of muscle synergies extracted from targeted surface EMG for estimating force during multiple-degree-of-freedom (DoF) contractions involving the wrist and hand. For this purpose, muscle synergies were extracted from twelve forearm muscles from eight able-bodied subjects. The constrained isotonic force produced by the wrist and the hand during these contractions was recorded along multiple axes, each responsible for one DoF. The derived neural inputs were then input into an artificial neural network (ANN) to estimate the force. The results were evaluated by comparing them with those obtained using mean absolute values (MAVs) for force estimation. The results obtained using muscle synergies were significantly better (p < 0.05) than those obtained using MAVs in the estimation of force when training with both 1- and 2-DoF contractions (p = 0.02) and also when training with only 1-DoF contractions (p = 0.001). The latter case was important, as a training protocol that includes all desired 2-DoF contractions is very difficult for amputee users. For this case, the results obtained using muscle synergies were significantly improved compared to those obtained using MAVs. In addition, the robustness of muscle synergies was examined across different force levels. The results indicate that muscle synergies are robust and reliable for the force estimation of multiple-DoF tasks, and are thus a promising approach for the proportional control of prostheses.