The effect of methadone dose on the cigarette smoking of five methadone maintenance subjects was studied in a clinical setting. Following a two-week baseline period, daily doses of methadone were either increased (50-80 mg) or decreased (80-50 mg) every two weeks according to an A-B-A-B study design. Continuous self-monitoring was used to collect data on natural smoking behavior, and expired air carbon monoxide (CO) levels were measured at each clinic visit. The predicted methadone dose-related changes in smoking were found in three of the five subjects. Fine-grained analysis of self-monitoring records showed that the proportion of total daily smoking was highest within 4 h after taking methadone for three subjects. CO levels were not significantly associated with smoking rate. The results further support and extend previous reports that methadone may produce dose-related increases in smoking. No reactive effects of self-monitoring were observed, and compliance with this procedure was extremely high, supporting the usefulness of this method for assessing natural smoking behavior in this population.