We studied socioeconomic status and marital status as predictors of smoking cessation, adjusting for previous smoking behavior and family background by using a large Finnish prospective twin dataset unselected for smoking behavior. The data were collected by postal surveys in 1981 and 1990, and the sample comprised 3,069 current smokers, of whom 20% had quit smoking by 1990. Logistic regression analyses of all twin individuals and conditional logistic regression analysis of discordant pairs were used to predict smoking cessation. High education predicted smoking cessation among both men (OR=2.32, 95% CI=1.31-4.10) and women (OR=3.98, 95% CI=1.85-8.51) as did high social class among women. Additionally, starting at a late age, smoking a small number of cigarettes per day, and a low level of nicotine per cigarette predicted cessation. Socioeconomic differences in cessation diminished only slightly when we adjusted for smoking behavior factors. Among the twin pairs who were discordant in terms of smoking cessation, the twin who continued smoking also smoked more on average at baseline (men: OR=.94, 95% CI=.89-.99; women: OR=.82, 95% CI=.71-.94). The male twins who continued smoking had a smaller probability of getting married during the follow-up than had the cotwin who had quit smoking (OR=3.91, 95% CI=1.02-15.02). Indicators of socioeconomic status were important predictors of smoking cessation even when we adjusted for previous smoking behavior. For men, marriage was associated with an increased probability of cessation