We often make decisions while moving, such as where to pass a ball while running in soccer. Perceptual decision-making paradigms have focused on how noisy goal-related evidence influences the evolving decision-making process, but not the subsequent movement. Sensorimotor control has examined motor planning prior to an abrupt indication of a correct target, but not the role of an evolving decision-making process on movement. We know little on the interplay between decision-making and movement control. Here we test the idea that the sensorimotor system uses a control policy that integrates and acts upon the evolving decision-making process prior to a decision. To investigate we extended upon the `tokens' task (Cisek, 2009) where participants reach forward to one of two potential targets. Once participants left the start position, 15 tokens provided goal-related evidence by individually moving into one of the two targets. We manipulated both the token patterns to influence the evolving decision-making process. Once participants felt confident which target would contain the most tokens, they were instructed to simultaneously hit that target with their cursor and push a button with their nondominant hand to indicate decision time. Results suggest that lateral hand movements are biased by the evolving decision-making process prior to a decision. An optimal feedback controller that uses a drift-diffusion model to select a target captures online reaching behaviour. Collectively, our findings support the idea that that the sensorimotor system uses an online control policy that integrates and acts upon the evolving decision-making process prior to a decision.