In an experimental study designed to investigate a decision‐making model of seat‐belt use, 227 employees of an agrochemical company participated in a health information program in which they watched either a videotape on seat belts or a control videotape and completed questionnaires immediately afterward and at 3 months and 1 year after exposure. In terms of total effects, the seat‐belt videotape influenced beliefs, fear, and intentions assessed immediately after exposure, but had no effect on self‐reported frequency of belt use at 3 months or 1 year. A full path analysis indicated some support for the decision‐making model. In particular, probability difference (the perceived reduction in risk of death or serious injury due to wearing a belt) had a large influence on intentions to wear a belt and partly mediated the effect of the videotape on intentions. Reported frequency of belt use at 3 months was influenced both by post‐test intentions and by initial frequency of belt use. Similarly, belt use at 1 year was affected by belt use at 3 months and by initial belt use. The findings are discussed in terms of the role of subjective probabilities and habitual factors in seat‐belt use.