AIMS: To measure reduction in exposure to smoke in two population-based studies of self-reported smoking reduction not using nicotine replacement. DESIGN: Cross-sectional analyses of biomarkers and smoking. SETTING: Data from two time-points in the Copenhagen City Heart Study (CCHS), 1981/83 and 1991/94, and the Copenhagen Male Study (CMS) in 1976 and 1985/86, respectively. PARTICIPANTS: There were 3026 adults who were smokers at both time-points in the CCHS and 1319 men smoking at both time-points in the CMS. MEASUREMENTS: Smoking status and tobacco consumption were assessed by self-completion questionnaire. Measurements of biomarkers of smoke intake were taken at the second time-point in the two studies: expired-air carbon monoxide (CO) in the CCHS and serum cotinine in the CMS. Biomarker levels in medium (15-29 g tobacco/day) and heavy (> 30 g/day) smokers at the first time-point who later reported a reduction in cigarettes per day of 50% or more without quitting were compared with continuing medium, heavy and light smokers (1-14 g/day) using linear regression. Sex (CCHS only), age, self-reported inhalation of smoke, duration of smoking, type of tobacco and amount smoked were included as covariates in multivariate models. FINDINGS: Heavy smokers who reduced did not show lower levels of biomarkers at follow-up. Medium smokers who reduced showed a reduction in cotinine but not CO. The reduction in cotinine was not commensurate with the reported reduction in consumption. CONCLUSIONS: Long-term reductions in self-reported tobacco smoking are probably associated with, at best, modest reductions in smoke exposure.