A set of television messages based on the findings of a preliminary study of factors associated with actually observed use of safety belts was produced and shown on one cable of a dual cable television system for nine months. Observations of actual use of safety belts were obtained before and throughout the study period. The automobiles in which safety belt use was observed were matched to the households of experimental and control cables through license numbers, motor vehicle registry files and the cable company files. The television messages had no effect whatsoever on safety belt use. This study adds to the growing body of evidence that a behavior modification approach is an inefficient and often ineffective means of reducing highway losses. Passive approaches, i.e., those which reduce the frequency and severity of damage to people and property irrespective of voluntary action show greater promise of reducing highway losses. (A)