Evaluation of the ability of cerebrally injured patients to return to driving is an important task for rehabilitation specialists. These evaluations require predictively valid methods of assessment based on identification of relevant skills and abilities. The present study tested a hypothetical model for driving after cerebral injury and determined its use in evaluating fitness to drive. Thirty-five patients with cerebral damage due to head injury or cerebrovascular accident participated in the study. All were administered (a) a predriver evaluation, that is, a battery of neuropsychological tests chosen a priori to test the model, (b) a simulator evaluation, and (c) a behind-the-wheel evaluation consisting of driving on a protected course and in traffic. The results showed that 93% of the driving outcome in traffic was explained cumulatively by findings from the predriver and simulator evaluations as well as from behavioral and operational measures during evaluation on the protected lot. These results supported the predictive validity of the model and are discussed in terms of methodology for evaluation of return to driving.