A basic tenet of empirical research on cigarette smoking behavior is the systematic assessment of patterns of use. However, the large majority of extant research relies on smokers’ retrospective reports of their average number of cigarettes per day (CPD), a measure that may be variable in terms of reliability and validity. Using data from 3 previously published studies of non-treatment-seeking daily smokers (combined N = 89), this analysis examined the reliability of self-reported CPD, the consistency of returned cigarette butts each day over 4 consecutive 24-hr periods, the validity of self-reported CPD compared with returned cigarette butts, and the relationship of CPD and returned cigarette butts to toxicant exposure. Results showed that self-reported CPD was reliable across telephone and in-person screening interviews (r = .87, p < .01). Although average self-reported CPD and returned cigarette butt counts did not differ significantly, t(87) = −1.5 to 0.3, all ns, butt counts revealed a wider range of variability in daily smoking behavior. In addition, self-reported cigarette use exhibited substantial digit bias (Whipple’s index = 413.8), meaning that participants tended to round their estimates to values ending in 0 or 5. Cigarette butt counts, but not self-reported CPD, were significantly associated with exposure to smoke toxicants. However, this former relationship was revealed to be linear, but not curvilinear, in nature. These findings have implications for both research and treatment efforts, as researchers often rely on accurate assessment of CPD to predict a variety of smoking-related outcomes.