BACKGROUND:People with Alzheimer's disease may be required to undertake clinical and on-road assessments to determine fitness to drive. The manner in which on-road assessments are conducted with drivers who do and do not have navigational problems may affect the outcome. OBJECTIVES:Investigate the effect of 1) navigational difficulties, 2) location of assessment (un/familiar area) and assessment order, and 3) undertaking a second assessment (practice), on passing an on-road driving assessment. METHODS:Forty-three drivers undertook an Occupational Therapy-Driver Assessment Off Road Assessment (OT-DORA) Battery which included the Drive Home Maze Test (DHMT). Participants with/without a history of navigational problems were randomly allocated into three groups: 1) Unfamiliar/then familiar area assessment; 2) Unfamiliar/unfamiliar; 3) familiar/unfamiliar. An on-road assessment protocol was used including over 100 expected behaviors at nominated points along the directed route. For familiar area assessments, the driver self-navigated from their home to shops and services. A pass/fail decision was made for each assessment. RESULTS:A generalized linear mixed effects model showed neither location, nor practice affected passing the on-road assessment. Participants with navigational problems were six times less likely to pass regardless of route familiarity and direction method, and the DHMT was a significant negative predictor of passing. CONCLUSION:Drivers with Alzheimer's disease who have navigational problems and are slow to complete the DHMT are unlikely to pass an on-road assessment. However, navigation and maze completion skills may be a proxy for an underlying cognitive skill underpinning driving performance.