Background. Thirty-two worksites participating in a randomized trial of worksite health promotion aimed at reducing smoking and obesity were categorized at baseline and 2 years later as having either restrictive or unrestrictive smoking policies. Between the two assessment points, 16 sites received health promotion interventions. Results. At baseline 15 sites had restrictive policies and 17 unrestrictive policies. Smoking restrictions were associated with significantly lower smoking prevalence and higher lifetime quit rates among ever smokers. They also were associated with more recent quit attempts and lower daily cigarette consumption, although these effects were not significant. Between baseline and follow-up, 9 of the 17 worksites that had few smoking restrictions at baseline became restrictive. Although neither baseline smoking policies nor changes in smoking policy predicted change in smoking prevalence or in the frequency of quit attempts, smokers in sites changing from unrestrictive to restrictive policies reported a significant reduction in daily cigarette consumption. Conclusions. The worksite health promotion program was successful in reducing smoking prevalence in intervention sites compared to controls. However, the existence of restrictive smoking policies neither helped nor hindered these intervention efforts. The present data are believed to support the idea that restrictive smoking policies have beneficial effects on the smoking habits of employees, but that the magnitude of this effect is modest.