Rehabilitation specialists, particularly occupational therapists, are often involved in evaluating clients' fitness to resume driving after cerebral damage due to head injury or stroke. Their available methods include separate or combined administration of predriver, simulator, and behind-the-wheel evaluations. However, use of these methods without a theoretical model on which to base test selection has yielded some criticism and little research about the effectiveness of these methods in predicting driving performance. Using the authors' Cybernetic Model of Driving as the basis for assessing abilities and behaviors relevant to driving outcome, this study sought to determine the effectiveness of the evaluation methods by discriminant analysis and measurements of sensitivity in predicting behind-the-wheel failures. Comprehensive evaluations of 106 patients revealed that residual deficits in cognition per se did not render a person unfit to drive and underscored the importance of considering behaviors in determining fitness. The methods of evaluation were shown to be relatively sensitive in predicting outcome; off-road and on-road evaluation reached sensitivities of 90% and 92% with the inclusion of behavioral measures. Formulae for predicting outcome based on methods of evaluation are provided and reasons for failures in behind-the-wheel evaluations are discussed.