This paper presents results of a "work in progress" project, addressing a research objective that seeks to create a deeper understanding of the on-road performance of drivers with cognitive impairment. The method is a quantitative statistical analysis of performance and error outcomes. Data has been sourced from a prospective study of the occupational therapy set route on-road assessment results of114 people. The data has been divided into four outcome groups; route completed / pass, route completed / fail, non- completed / pass and non- completed / fail. Additionally the outcome scores have been clustered into subsets of low, moderate and high density for each individual drive. Outcomes have been compared to a number of variables including level of cognitive deficit, age and gender. The route has been coded and defined in regard to task elements and complexity. The observed and rated human performance elements of each driving task have been allocated a nominal code. The number of errors committed per driver have been quantified and given a raw score, with subsequent calculations of differences between groups. The error types have been thematically clustered, e.g. hazardous or scanning error. By entering the data into a computerised statistical package it has been possible to objectively identify which driving tasks the drivers were most often able to perform consistently or safely and those in which performance was rated by an occupational therapist and a driving instructor as absent or unsafe. Preliminary analysis from this study has identified interesting results in a number of scenarios that are strongly associated with some published findings in the field of accident research. This research is being conducted as an adjunct to the now completed SA Repatriation General Hospital, Memory Disorders Study Unit Dementia and Driving Clinical Trial previously presented at the Road Safety Conference in 2000.