The decision to return to driving is both complex and difficult. It is often made with clinician support, perhaps on the basis of results from standardized paper and pencil tests, and less frequently an assessment of on-road driving. However, the resources required and inherent reliability and validity challenges suggest that greater use of computerization and driving simulation may play a useful role in the assessment process. In this study, thirty-six age-matched healthy and post-stroke drivers completed standard psychometric assessments (including NART, MMSE, BADS, IADL), computer delivered cognitive assessments (including SART, N-Back, Simple Reaction Time), as well as simulated and on-road assessments of driving. While significantly different in terms of psychometric and computer-based assessments, the healthy and post-stroke drivers who completed the on-road test did not differ. In contrast, driving in the portable simulator showed impaired driving, on some but not all driving manoeuvres, for those who failed the on-road test or had voluntarily ceased to drive. These results are discussed in terms of the implications they have for the multi-faceted nature of driving as a skill, and the need to involve simulation when assessing fitness to drive.