This paper introduces an approach for dynamically adapting the level of automation of a wheelchair system on the basis of the current level of the user's motor abilities. For this purpose, a study is described, during which participants drove through a course consisting of 14 sub-sections. The predictive validity of the participants' wheelchair driving behavior on their precision, tremor, wrist-finger velocity, arm-hand velocity and aiming capability was analyzed. The results demonstrate impressively (1) that some course sections can better be used in order to derive variables allowing a reasoning on the users' motor ability level than other sections and (2) that especially the precision ability can be predicted e.g. on the basis of the distances driven in the forward mode. Implications on how an adaptive assistance system for powered wheelchair could be implemented are discussed as is the impact of such a system on the wheelchair market.