The aim of the study was to examine whether smoking reduction over a 6-year period (1975-1981) predicted smoking cessation 9 years later by 1990. The longitudinal data from three surveys over 15-year period among the Finnish adult twin cohort were used. The data were collected by postal surveys in 1975, 1981, and 1990, and the sample comprised 11,015 participants, of whom 2,443 were persistent current smokers in 1975 and 1981. Smoking cessation by 1990 was examined as the outcome measure. Nineteen percent reported having stopped smoking by 1990, corresponding to an approximate annual rate of smoking cessation of 2%. Those who had quit smoking by 1990 had larger decreases in smoking between 1975 and 1981. The odds ratio for quitting by 1990 increased with increasing levels of reduction in amount smoked between 1975 and 1981. This was found in both categorical and trend analyses, and when adjusting for age, sex and for amount of baseline smoking. The model among twin pairs discordant for cessation indicated that the association of smoking reduction with increased likelihood of cessation was independent of genetic or early shared family influences on smoking behavior. We conclude that smokers who are able to decrease the number of cigarettes smoked per day by at least 25% are more likely to quit later.