This review suggests several hypotheses about the cognitive developmental mechanisms involved in the motor deficits of children with cerebral palsy. We suggest a new theory that visuospatial deficits involving the manipulation of multiple spatial reference frames are crucial components of the disorder in spatial orientation, manipulation, locomotion, navigation, and even social interactions. We review basic knowledge about the brain networks involved in spatial memory and cognition. We then present several potential paradigms for studying specific deficits. We consider first the use of vestibular signals for egocentric spatial orientation in children and the 'locomotor trajectory paradigm' for studying gaze anticipation and perceptual components of walking. We then describe new paradigms for studying egocentric and allocentric strategies in spatial tasks: the 'virtual path length', the 'virtual palace' and the 'virtual star maze'. We also consider paradigms involving the use of other persons and perspective change from a first person's to a third person's viewpoint as reference in spatial tasks or social interactions: the 'designation' paradigm, the 'harlequin', and the 'tightrope walker'. Finally, we briefly present a new experimental set up involving a 'virtual carpet', which follows previous studies of cognitive strategies for generating locomotor trajectories using the 'magic carpet' and which will allow a large variety of studies involving executive functions and inhibition of the first-appearing strategies during development. Several of these new paradigms could be used for remediation.Children with cerebral palsy (CP) suffer from several 'motor' or 'sensory-motor' deficits in tasks involving spatial cognition.1 These deficits can be observed either in spatial orientation, in manipulation, or in locomotion. Most of the available neuropsychological tests about this problem are table tests, or even simple paper tests. However, what is required are tests exploring the competences of these patients in more ecological situations and in relation to the cognitive brain functions involved for visuospatial tasks in proximal but also in remote environments, requiring navigation, locomotion, and use of multiple scales of space from the nearest to the environment.In this review we shall first state some general hypotheses proposed by one of us (AB) about the cognitive visuospatial and spatial memory deficits that may be involved in the observed 'motor' deficits of children with CP. We shall then review a few potential new paradigms for studying these deficits, and potentially their remediation, and briefly describe a new version of the 'magic carpet' as proposed by one of us (MZ). The purpose of this paper is therefore not to give a report of each study, whose details can be found in the cited papers, but to provide a conceptual and practical framework of the understanding of these deficits from childhood to adulthood and potentially in elderly patients. The work is essentially based on that done by our laboratory or ...