Although cigarette smoking is the strongest known risk factor for lung cancer, the effects of specific smoking practices have not been completely characterized. The present study examines determinants of lung cancer risk in a population-based, case-control study conducted in New Mexico, 1980-82. The study included 521 cases and 769 controls matched for age, sex, and ethnicity. Either the index subjects or their next-of-kin were interviewed in person to obtain a detailed history of cigarette smoking and information concerning other risk factors. With the use of multiple logistic regression, a model was constructed of the effects of amount smoked, duration of smoking, cigarette type, and smoking cessation on lung cancer risk. Among current smokers, risk increased with each additional cigarette smoked per day (P less than .001). For duration of smoking, the risk per year smoked in individuals 65 years and older was only one-third that in persons under age 65 years. With regard to cigarette type, a somewhat higher risk was found associated with smoking nonfilter cigarettes, but there was no evidence of decreasing risk as the extent of filter smoking increased. Lifelong filter cigarette smokers and smokers of both filter and nonfilter cigarettes were at lower risk than lifelong smokers of nonfilter cigarettes only. In ex-smokers, the pattern of variation of relative risk with amount and duration was similar to that in the current smokers. Excluding those who had stopped for 1 year or less, the relative risk declined exponentially with duration of smoking cessation (P less than .01). These analyses confirm the strong benefits of smoking cessation and indicate possible reduction of risk from smoking filter cigarettes.