A follow-up of smoking behaviour to age 18 in a longitudinal study of a birth cohort enabled an assessment of the prevalence of smoking and quitting among adolescents approaching adulthood. There was a dramatic increase in number of daily smokers (15% at age 15 years to 31% at age 18 years), and in a climate of expected decreases in smoking, a history of never smoking to age 15 years was not as protective against future smoking as anticipated. Among 15-year-olds who had experimented with smoking, only 11% per year stopped by age 18 years. Cessation rates for adolescent daily smokers were low (3% had not smoked in the last year at age 18 years), and previously have not been widely reported. We also examined some methodological issues related to self-reported prevalence rates, in particular reliability, validity and sources of bias, finding confirmation of the accuracy of information from prospective longitudinal studies and supporting the conclusion that adolescents' recall for information beyond a 1-year period is inconsistent.