The study examines the role of several potential predictors of urinary cotinine levels in a cross-sectional sample of 1,072 nonsmoking children and adolescents in Latium, Italy, during 1990-1991. As expected, there was a strong relation between passive exposure to smoking and the amount of maternal and paternal self-reported smoking. The urinary cotinine level increased with a decreasing level of paternal education and with an increasing index of household crowding; self-report of recent exposure to smoking outside the home was a strong predictor of the biologic marker. The analysis was then restricted to 346 subjects whose parents claimed that they were nonsmokers and that there were no smokers at home. In this group, however, 57 children reported some active smoking at home by their parents. Those with parents suspected to be "deceivers" had higher level of urinary cotinine than did those truly not exposed. In addition, urinary cotinine in this group was clearly associated with duration of exposure to smoking outside home. The study indicates that both factors related to family circumstances and exposure outside the household setting are strong determinants of urinary cotinine levels. The finding may be considered a direct confirmation that passive smoking among children should be viewed as a specific community responsibility.