We studied the association between self-reported exposure to passive smoking (ETS) and urinary cotinine among 95 ex-smokers, who had quit in the context of a smoking cessation trial in general practice, and among 33 ex-smokers who had been contacted in the context of a study aimed at estimating the spontaneous cessation rate among smokers, listed in the roosters of the General Practitioners participating in the trial, who had not been offered recruitment. The results of multivariate linear regression analysis indicated that duration of exposure at home was strongly related to cotinine levels (increase of 1 ng/mg every 3 hours of exposure at home), quantified by radioimmunoassay. Duration of exposure in public places showed a similar effect, but the association was significant only among subjects who reported having been exposed in a heavily ETS polluted environment (increase of 1 ng/mg every 45 minutes of exposure). Among subjects exposed at work cotinine concentration was associated to subjects' judgement of the concentration of ETS, rather than to duration of exposure. Standardized questionnaires may provide a reasonably accurate description of ETS exposure and they can be used to assess exposure in etiologic studies. Non-smokers non exposed to ETS at home may not represent an appropriate control group in such studies. Indeed, they may be exposed to other sources, as indicated by the relatively high levels of urinary cotinine among ex-smokers non exposed at home in our study. Misclassification of these subjects may result in a dilution of the effect of the exposure.