Effects of modeling on car safety belt use were investigated in a field experiment. Modeling, anticipated trip length, and gender of the model were manipulated in a 2 × 2 × 2 factorial design. Sixty-four female college students were told that they were to participate in an experiment that would take place in another location, requiring a drive either of less than one mile or of several miles. The driver either used or did not use a safety belt. Subjects' belt use was significantly related to the model's behavior. When the driver used a safety belt, 77.4% of the subjects used one; when the driver did not use a safety belt, only 313% of the subjects used one. A significant effect for trip length was also found. In the long trip condition, 71.9% of the subjects used a belt; in the short trip condition, 35.5% used one. Results are interpreted as support for a social learning theory approach to increasing voluntary safety belt use.