The purpose of the study was to determine the feasibility of developing procedures to measure driver decision-making performance and the feasibility of improving that decision- making performance through training. The study consisted of five interrelated tasks. Critical driving situations were identified and driver decision-making requirements were analyzed within a human information-processing framework. A review of measurement devices and training techniques revealed serious deficiencies with respect to decision making. Four specific dimensions were identified as related to effective decision-making. A classification-training model was developed and related to a probabilistic policy making cost-benefit model. A comprehensive research and development plan is proposed for four separate groups of drivers.