Both genetic and environmental factors affect smoking initiation and maintenance, but less is known about the genetic architecture of various other smoking-related behaviors. The aim of this study is to examine the genetic architecture of smoking behavior in a large twin cohort. Questionnaires with an extensive smoking history section were mailed to same-sex adult twins of the Finnish twin cohort. The final study population included 2923 monozygotic and 6018 dizygotic twin pairs aged 24 to 88 years. Two-stage bivariate genetic modeling of age at initiation with amount smoked (less than 20 cigarettes per day vs. 20 or more) and age at initiation with smoking cessation was done by using the Mx statistical package. For men the heritability estimate for age at initiation was .59 (95% confidence interval [CI] .49-.69), for amount smoked .54 (95% CI .45-62) and for smoking cessation .58 (95% CI .50-.65). For women the heritability estimates were .36 (95% CI .2S-.43), .61 (95% CI .46-.70) and .50 (95% CI .39-.60), respectively. The genetic correlations between age at initiation and amount smoked or smoking cessation were at most .22 in magnitude, indicating that genetic influences in age at initiation accounted for at most about 4% of the genetic factors in amount smoked or in cessation. Genetic factors are important in amount smoked and smoking cessation and they are largely independent of genetic influences on age at initiation. This has implications for defining phenotypes in searches for genes underlying smoking behaviors.