The importance of reassessing the psychometric properties of an existing scale is highlighted. The construct validity of a driver attitude scale was not confirmed by factor analysis and the original scale scores were shown to have poor to moderate internal reliability. On the basis of a re-analysis of data from 1,059 repeat traffic offenders, 39 of the original 47 profile items were retained to form a total score. The computation of a total profile score for these items resulted in marked improvement of the profile's internal reliability (alpha=.89). Total scores were significantly different based on sex (p<.01), age, and education level of these drivers (p<.001). Total profile scores were significantly related to past accident and violation experience, with negative attitudes increasing as the number of self-reported incidents decreased (p<.001). Instructional and evaluation implications are discussed. (Contains 1 table and 12 references.) (Author/SLD)