The stability of plasma cotinine was studied in smokers and the validity of a brief questionnaire on environmental tobacco smoke (ETS) exposure was assessed in nonsmokers. The correlation between two cotinine levels measured 14 weeks apart was 0.81 in smokers (n = 148). A single measurement will classify 95% of all subjects within 30% of their habitual plasma cotinine level. Plasma cotinine was higher in ETS-exposed nonsmokers (n = 50) than in ETS-unexposed nonsmokers (n = 55; 1.6 ng/ml vs 0.6 ng/ml; P < 0.0001), but there was a substantial part of intersubject overlap. The sensitivity and specificity with respect to self-reported smoke exposure were 56 and 89%, respectively. This study shows that a single plasma cotinine determination gives a good impression of cotinine levels. It is dubious if a brief questionnaire about hours of passive smoke exposure in nonsmokers is valid to distinguish ETS-exposed from ETS-unexposed nonsmokers.