The aim of this thesis is to use the cue-integration framework as a normative model to study the role of binocular depth cues in grasping, and to explore how the visuomotor system decreases uncertainty in visual information. We explored whether the availability of different cue types, or visual uncertainty per se determines the margin-for-error in programmed grip apertures. Results showed that both monocular and binocular cues contributed to grasping performance, and that performance was improved when both cues were available compared to when either cue was available alone. We also examined whether cue type or the precision of feedback determines online grasping performance. Our findings suggested that the contribution of feedback from binocular and monocular cues was determined by the precision of information in a particular situation. Taken together, the above results suggested that the idea of a binocular specialism for grasping is incorrect, and that a cue-integration account better explains the data. We extended this idea to examine whether learnt, familiar size information is used alongside visual depth cues. The results suggested that familiar size is also integrated with binocular and monocular depth cues. The above experiments confirmed that the visuomotor system opens the grasp wider when visual uncertainty increases in order to increase the margin-of-error in the movement. We examined this behaviour further by exploring different movements, for which a different response to perceptual uncertainty is required. We found that the visuomotor system encodes both perceptual uncertainty, and the probability of making an error to programme grasping movements. Overall our results suggest that binocular information is not critical for the planning or online control of grasping. Rather, our findings suggest that all sources of information are integrated for grasp control, minimising uncertainty in estimates of object properties, which allows for smaller margin-of-error in grasping movements.