The aims of this study were to determine the effects of smoking on the incidence and prognosis of erectile dysfunction (ED) and that of ED on smoking behavior, and to find out whether smoking either directly or through vascular diseases causes ED. The target population comprised of all men born in 1924, 1934 or 1944 and residing in the city of Tampere or 11 adjacent municipalities in Finland. Questionnaires were mailed to 3,143 men in 1994, to 2,864 in 1999 and to 2,510 men in 2004. The response rates were 70%, 75% and 75% respectively. ED was assessed using two questions on subjects’ erectile capacity. Logistic and Poisson regression models were used in the multivariable analyses. Current smoking was associated with ED (Adjusted prevalence odds ratio (POR) = 1.7, 95% CI 1.2-2.4), but not with ex smoking. The incidence of ED non-significantly increased (incidence density ratio (IDR)=1.4) and that of ED recovery reduced (IDR=0.7) with current smoking. A dose-response relationship was found between smoking and ED. Although the relative risks estimates were not statistically significant, probably due to small numbers. Only heavy smokers were significantly at higher risk of ED. Compared with non-smokers, confounder-adjusted IDR was 1.6 (95% CI 1.0-2.6) for men who smoked 21 cigarettes or more daily. Both quitting (IDR=1.7) and starting smoking (IDR=1.9) were rare and higher in men with ED. However, the IDRs estimates were not statistically significant. Current smokers at baseline (1994) who developed vascular disease during 1994-1999 were three times (Confounder-adjusted IDR=3.1, 95% CI 1.3-7.5) at higher risk of ED during 1999-2004 compared with never or ex smokers who did not develop vascular diseases. On the other hand, current smokers who did not develop vascular diseases were not at higher risk of ED (IDR=1.0). There were two bi-directional relations between smoking and ED. Smoking caused ED though vascular diseases and ED caused smoking. The recovery from ED was less in smokers than in non-smokers, and current smokers with ED were more likely to stop smoking compared with men free from ED.