Meta-analytic techniques were applied to 20 controlled studies of work site smoking cessation yielding a total of 34 comparisons of long-term (average = 12 months) quit rate (QR). An overall weighted mean effect size (ES) of .21 +/- .07 was found, indicating a modest but significant overall effect (P less than .01). The weighted average follow-up QR from all interventions was 13%. Based on previous research, characteristics associated with interventions, work sites, employees, and research methodology were identified as potential moderator variables. Apart from methodological variables, interventions conducted in smaller work sites (ES = .45 +/- .17), which lasted 2 to 6 hours (ES = .42 +/- .13), and which contained heavy smokers (ES = .28 +/- .07) were associated with the largest effect sizes. We were also interested in absolute quit rates. After controlling for methodological quality, programs that included a cessation group component (partial r = .39), that were not overly complicated (partial r = -.42), and that shared company and employee time (partial r = -.48), as well as the above variables had the strongest associations with QR. Implications for public health policy and future research are discussed.