The accuracy and reliability of saliva cotinine as an objective measure of smoking status was examined in two field studies. In Study I, saliva was collected from smokers and nonsmokers with repeated samples taken from a randomly selected subset of the smokers. Results indicated perfect classification of smokers versus nonsmokers and acceptable reliability of repeated samples. Study II investigated the accuracy of saliva cotinine in detecting recent quitters in a worksite smoking cessation program. Saliva cotinine showed greater accuracy than expired carbon monoxide at detecting quitters, provided they were abstinent for at least seven days. From pre- to post-treatment, subject's saliva cotinine levels dropped 19 per cent while self-reported rate of smoking dropped 54 per cent. Saliva collection in the field is feasible and cotinine appears to be one of the more sensitive assays currently available for epidemiologic and clinical applications.