OBJECTIVE: To examine outcomes and predictors of smoking cessation among elderly patients treated for nicotine dependence. DESIGN: Retrospective analysis of patients aged 65-82 who received a nicotine dependence consultation at the Mayo Medical Center between 1 April 1988 and 30 May 1992. Patients were contacted by telephone by a trained interviewer six months after the consultation and were sent a follow-up survey in August 1993. SETTING: Mayo Medical Center, Rochester, Minnesota, United States. SUBJECTS: A total of 613 patients (310 men, 303 women) with a mean age of 69.0 (SD 3.5) years were seen during the study period. MAIN OUTCOME MEASURES: Point prevalence self-reported smoking status. Patients were considered abstinent if they self- reported not smoking (not even a puff) during the seven days before contact. RESULTS: At six-month follow up, 24.8% of the 613 patients reported abstinence from smoking. On multivariate analysis, smoking abstinence was more likely if patients were hospitalised at the time of the consultation, married to a non-smoking spouse, very motivated to stop smoking, and reported their longest time of previous abstinence to be less than a day or more than a month. The response rate to the mailed follow-up survey was 69.9% (429 of 613). The mean duration of follow up was 40.0 +/- 13.2 months following the consultation. Of the 429 patients, 103 (24.0%) reported abstinence from smoking and 326 (76.0%) were smoking at six-month follow up. Patients who reported abstinence at six months had a higher cessation rate at the last follow up (76.0%) compared with patients who were smoking at six-month follow up (33.0%, P < 0.001). For patients who were not smoking at six months, no factors were found to significantly predict abstinence at last follow up. For patients who were smoking at six months, factors associated with smoking cessation at last follow up were: more than a year as the longest time off cigarettes before the consultation; counsellor rating of less severe nicotine dependence; and older age at first regular smoking. CONCLUSIONS: Several predictors of smoking cessation were identified in this study which may be useful for tailoring smoking interventions for the elderly.