The interrelationships among education, smoking, and non-cardiovascular (non-CVD) mortality were examined in middle-aged white males from the Chicago Peoples Gas Company Study (PG), the Chicago Western Electric Company Study (WE), and the Chicago Heart Association Detection Project in Industry (CHA). In each study, college graduates had the lowest prevalence of current smokers and the highest prevalence of former smokers. The associations between education and smoking were strongest in CHA, a study with baseline measurements 10-14 years after those of PG and WE and 3-8 years after the US Surgeon General's report on smoking and health in 1964. In PG and WE, the relative risks of non-CVD death for those who did not attend college compared to those who did were 1.50 and 1.38 (95% limits, 1.04 to 2.18 and 0.95 to 2.02). In CHA, the relative risk for those who did not graduate from college compared to those who did was 1.55 (1.17, 2.05). Differences in baseline cigarette smoking could account for only 23-29% of these increased risks. Because the associations between education and non-CVD mortality may have been confounded by changes in smoking status over the course of follow-up in these studies, non-CVD deaths were subdivided into those from causes related to smoking and causes not related to smoking. For smoking-related causes, the relative risk of death for those who did not attend/graduate from college was 1.95 (0.96, 3.95) in WE, 2.13 (1.18, 3.87) in PG, and 2.34 (1.47, 3.84) in CHA, while the relative risks for causes not related to smoking were 1.17, 1.12 and 1.16, respectively. These findings suggest that education is related inversely to non-CVD mortality primarily through smoking and smoking-related causes of death. With smoking becoming increasingly a habit of the less well-educated, these findings underscore the need for smoking prevention and cessation programmes targeted at the lower end of the socioeconomic scale.