In this investigation participants formulated a grip aperture (GA) consistent with the size of an object embedded within a Muller-Lyer (ML) figure prior to initiating visually guided grasping movements. The accuracy of the grasping response was emphasized to determine whether or not the visuomotor system might resolve the premovement bias in GA very early in the response, as predicted by the perception/action model (PAM: Milner & Goodale, 1995), or more gradually as the action unfolds over time, as predicted by the planning/control model (PCM: Glover & Dixon, 2002). Grasping time (GT) was normalized and GA was measured at 11 time points: a premovement GA and ten time points beginning at 10% of GT and ending at 100% of GT. Not surprisingly, premovement GA was influenced in a direction consistent with the perceptual effects of the ML figure (cf. Westwood, Heath, & Roy, 2000). Interestingly, however, the bias in GA was not attenuated during the response, that is, a reliable illusory effect was observed for upwards of 90% of GT. Only at 100% of GT did we find resolution of the illusory effect. Based on these results, it is proposed that shaping GA in advance of response execution leads to a feedforward mode of grasping control in which the spatiotemporal characteristics of GA are controlled via an offline and context-dependent planning mechanism (i.e., the ventral visual pathway).