To provide a scientific background in road safety domain a better understanding of human risk factor is crucial. The aims of the present study were the following: (1) developing an accident prediction model for estimating the at-fault accidents of drivers (2) controlling for the regression-to-the-mean and screening out the accident-prone drivers (3) identification of significant behavioral predictors in at-fault accident occurrences and delving into the relationship between the aberrant driving behaviors and at-fault accidents of those identified as accident-prone. A questionnaire survey compiling various measures of personality type, aberrant driving behavior, demographic and accident history information of 1762 Iranian drivers was conducted in which 1375 male and 387 female participants were of the average age of 35.6 (SD=11.987). To analyze the obtained data, the generalized linear modeling (GLM) approach was taken resulting in four models with various independent variables. The results indicated that age, gender, education level, years of active driving, and especially exposure had an effect on drivers’ at-fault accidents while there was no discernible effect from income level, personality type and area of residence. In the screening procedure, 715 drivers were identified as accident-prone. Behavioral comparison analyses indicated that the lapses, errors, ordinary and aggressive violations are different for the accident-prone drivers. A comparison between the accident-prone and non-accident-prone drivers revealed that the ordinary violations have considerably higher effect than the others on at-fault accidents. Implications of the results are discussed with regard to insurance policies and education interventions.

Keywords