Reaching movements made with hemiparetic upper limbs are often followed by compensatory trunk and shoulder-girdle movements, especially in reach-to-grasp strategies. We investigated the effect of restraint therapy on movement aiming at targets placed within arm reach. Eleven subjects were selected to participate in this study and were submitted to training with the trunk fixed to a chair during the reaching movement. Quantitative and qualitative scales (Modified Ashworth Scale, Fugl-Meyer Assessment Scale-upper limb section, Berg Balance Scale and Barthel Index) were used to evaluate reaching in three periods--pre-treatment, post-treatment and follow-up 3 months after training was completed. We found that in the arm section of the Fugl-Meyer Assessment Scale, pain (x2=8.22, P=0.016), proprioception (x2=6.00, P=0.049), flexor synergy (x2=6.07, P=0.048), wrist (x2=6.50, P=0.039), coordination velocity (x2=7.05, P=0.029) and the total score for the upper limb (x2=7.95, P=0.019) had statistical significance, especially in the pre-treatment and follow-up phases. The same happened using the Barthel Index (x2=8.33, P=0.016). Trunk restraint allowed patients with hemiparetic stroke to make use of active arm joint ranges that are present but not normally recruited during unrestrained arm reaching tasks. Appropriate treatments, such as trunk restraint, may be effective in uncovering latent movement patterns to maximize arm recovery in hemiparetic patients.