The evaluation of the ability of patients to return to driving after cerebral damage stands out as one of the most important tasks confronted by rehabilitation professionals. The present study was designed to critically assess evaluations that were developed at one facility to determine fitness to drive: an off-road, predriver evaluation of skills regarded as important in driving and an on-road, behind-the-wheel evaluation of abilities needed to drive in actual traffic situations. The evaluation results of 37 patients with cerebral damage due to traumatic head injury or cerebrovascular accident were studied retrospectively. Only 4 out of 21 items on the predriver evaluation significantly predicted the outcome of the predriver evaluation and none of the predriver evaluation items predicted the outcome of the behind-the-wheel evaluation. Only 6 of the 26 tasks on the behind-the-wheel evaluation significantly predicted the outcome of the behind-the-wheel evaluation. None of the items on the predriver evaluation or the behind-the-wheel evaluation explained a significant portion of variance related to outcome. The lack of internal and predictive validity of driver evaluations is discussed in light of these findings, and recommendations are given for improving the predictive power of driving evaluations.