Studies show that the proper design of clearance intervals has significant implications for intersection safety. For example, in 2001, approximately 218,000 red-light-running crashes occurred at signalized intersections in the United States. These crashes resulted in nearly 181,000 injuries and 880 fatalities and an economic loss of $14 billion. Driver behavior while the driver is approaching high-speed signalized intersections at the onset of a yellow indication varies as a function of many parameters. Some of these parameters are related to the driver's attributes, e.g., age, gender, perception-reaction time, and acceptable deceleration levels. Other parameters that relate to the intersection geometry include the approach speed, distance, and time to the intersection at the onset of the yellow indication. This study developed a novice approach for computing the clearance interval duration that explicitly accounts for the reliability of the design (probability that drivers are not caught in a dilemma zone). Lookup tables based on the limited data available from this study are provided to illustrate how the framework could be used in the design of yellow timings. The approach was developed using data gathered along Virginia's Smart Road test facility for dry and clear weather conditions for two approach speeds: 72.4 km/h (45 mph) and 88.5 km/h (55 mph). Each dataset includes a complete tracking of the vehicle every deci-second within 150 m (500 ft) before and after the intersection. A total of 3,454 stop-run records were gathered. These include 1,727 records (687 running records and 1,040 stopping records) for an approach speed of 45 mph and 1,727 records (625 running records and 1,102 stopping records) for an approach speed of 55 mph. Using these data, models that characterize driver perception-reaction times and deceleration levels were developed. The application of the proposed approach demonstrates that the current design procedures are consistent with a reliability level of 98%.