Using a complete factorial design, we tested three interventions for smoking cessation in routine primary care practice. The interventions tested were 1) physician counseling, 2) mailed letters and educational materials designed by the National Cancer Institute (NCI), and 3) referral to smoking cessation classes. Thirty-seven family practice physicians at three of Group Health's outpatient facilities participated. Patient participation rates were 95%, and follow- up was complete for 92% of those participating. None of the interventions had any effect on point prevalence of quitting as determined 8-9 months later by self-report. However, the combination of physician counseling and NCI materials doubled the odds of occurrence of significant antismoking behavior (quit, quit and relapse, or cut down) during the ensuing 8-9 months in those individuals receiving that combination. Referral to smoking cessation classes was strikingly ineffective in this setting. Of 369 individuals designated by study design for referral, only 14% even investigated the classes. This compares with a 10% self-referral rate for those persons not designated for referral by our study design. Our results and other recent work suggest that more intensive interventions on multiple occasions based on relapse prevention strategies hold promise for future success in smoking cessation efforts in primary care.