This study developed a self-assessment tool for predicting unsafe driving of senior drivers, based on a Rasch analysis. In order to develop this assessment, existing approved assessment items were collected and Delphi method with a panel of experts was used on elderly, driving and senior drivers, yielding 44 preliminary assessment items. Using these items, data of 339 senior drivers were collected and conducted a Rasch analysis. The person fit statistics was analysed and 31 subjects were found to be misfit. The data of the remaining 308 subjects were used to analyze the item fit statistics of the assessment; 6 items were judged to be misfit, and finally 38 items were selected. These items and the senior drivers were arranged based on logit-transformed difficulty and ability scores. The most difficult item was “You have difficulty reading the small print in the navigation or map.” with a logit score of 1.60, while the easiest item was “In the past 2 years, you have experienced dizziness, seizure or loss of consciousness.” with a logit score of -2.12. These items can be interpreted as the most common or rarest symptoms in senior drivers, respectively. In addition, the ability of the subjects was arranged from -2.44 to 7.32, and this was used to produce the final transformation formula for subject ability. Through rating scale analysis, a 3-point scale was established, consisting of 1 point (“Agree”), 2 points (“Disagree”), and 3 points (“Strongly disagree”). Using factor analysis, the following sub-areas were identified: “on the road, 9 items, 27 points,” “coping, 14 items, 42 points,” and “health, 15 items, 45 points,” making a total score out of 114 points across all items. Using a cut-off value of 75.5 points, this assessment was able to identify unsafe senior drivers with probabilities of 88.6% (positive) and 86.3% (negative).