In two studies, children between 4 and 6 years old were tested on a radial search task requiring nonredundant sampling of eight identically labeled positions in a room. In the training phase (5 days), children made free choices by walking between positions, were passively transported in a pushchair, actively directed their own route from a pushchair, or were led on foot to positions selected by the experimenter. When tested (whether walking or directing while seated in a pushchair), children who had either walked independently or directed the experimenter while being pushed performed competently; those led on foot without spatial choice performed almost as well. Only the children who had neither independent locomotor experience nor autonomous choice performed very poorly. The results are related to neurobiological models of spatial cognition and may have implications for the transportation of children with mobility problems.