We examined the relationship of cigarette tar yield and other cigarette-usage characteristics in current smokers to the incidence of lung cancer in a study population of 79,946 Kaiser Permanente Medical Care Program members, aged 30-89 years, who completed a detailed, self-administered, smoking-habit questionnaire during the years 1979 through 1985. Mean length of follow-up was 5.6 years. There were 302 incident lung cancers, of which 89 percent occurred in current or former smokers. The tar yield of the current cigarette brand was unassociated with lung cancer incidence (relative risk [RR] = 1.02 per 1 mg tar-yield in men, 95 percent confidence interval [CI] = 0.98-1.05; RR = 0.99, CI = 0.96-1.03 in women). However, in long-term (> 20 years) smokers, the risk of lung cancer was decreased in women who had smoked filtered cigarettes for 20 or more years relative to lifelong smokers of unfiltered cigarettes (RR = 0.36, CI = 0.18-0.75), but not in men who had smoked filtered cigarettes for 20 or more years (RR = 1.04, CI = 0.58-1.87).