Abstract The present study examines the contributions of vision for perception processes in action. To this end, the influence of allocentric information on different action components (i.e., the selection of an appropriate mode of action, the pre-planning and online control of movement kinematics) is assessed. Participants ( n = 10) were presented with a shaft of various lengths (i.e., 13–20 cm) that was embedded in a Müller-Lyer figure. Picking up the shaft would, dependent on its length, either require a one- or a two-handed grasp. In different conditions participants were instructed to give a verbal judgement on the size of the shaft (VSJ); to make a manual estimation of the shaft's length (MLE); to indicate verbally whether they would grasp the shaft with one- or two hands (VAE); to actually grasp the shaft (G). We found that the Müller-Lyer figure affected the choice between using a one- or two-handed grasp, both when the participants actually grasped (G) the object and when they made a verbal estimation (VAE). The illusionary bias was of a similar magnitude as the one found in the verbal (VSJ) and manual perception task (MLE). The illusion had only a minor influence on the movement kinematics, and appears to be restricted to participants in which the grasping condition was immediately preceded by the VSJ-condition. We conclude that vision for perception contributes to the selection of an action mode, and that its contributions beyond that stage are dependent on the particular (experimental) circumstances.