The aim of this pilot study was to determine the most sensitive urinary cotinine level able to assess environmental tobacco smoke (ETS) exposure. 54 Florentine subjects (29 males and 25 females), reporting to be nonsmokers and exposed (E) or not exposed (NE) to ETS at home, at work or in places of recreation, were examined. The urinary cotinine concentration was determined using gaschromatographic analysis in samples collected on three consecutive days. 18 subjects (33.3%) reported to be exposed to ETS had a greater median cotinine concentration than 36 ETS-NE subjects (E = 3.3 pg/L vs NE = 2.2 microg/L, median values), with borderline statistical significance (P = 0.05). The 2.5 microg/L cotinine concentration was the only statistically significant cut-off (P = 0.04) discriminating between ETS-E to ETS-NE subjects, identifying 51.9% of the subjects examined as exposed (E). Considering the expanded uncertainty of measurement of the method used (20%), urinary cotinine concentrations higher than 3.1 microg/L, a value whose confidence interval is higher than our proposed cut-off of 2.5 microg/L, mean that to be sure that a subject is exposed to ETS.