This prospective study of the elderly population estimated the risks of smoking for morbidity and mortality and identified whether cessation of smoking reduced the risk of disease. Data came from face-to-face interviews that used a population-based probability sample of those aged 60 years or over in Taiwan, provided by the Population and Health Research Center, Bureau of Health Promotion. In total, 4,049 subjects were included at the baseline year of 1989 and followed up in 1993 and 1996. Smoking-related variables included current smoking status, smoking history, daily consumption, and years since the cessation of smoking. Cox regression models were used to analyze the relative risks for morbidity and mortality, controlling for demographics, physical function, and comorbidities. The sample was made up of 50.2% nonsmokers, 15.2% ex-smokers, and 34.6% current smokers in the baseline year. Current smokers were more likely to have lower respiratory tract diseases throughout the study. Current smokers had a higher risk of stroke from 1989 to 1993. No dose-response relationship for smoking exposure or impact of years since smoking cessation was found. Whether cessation of smoking is protective should be investigated for middle-aged adults followed to old age. An effective strategy for smoking cessation in the elderly is suggested, and people should be encouraged to quit smoking at any time.