OBJECTIVE: The objective of this study was to determine the validity of tobacco questionnaires when using as gold standard either a single biomarker or a combination of two biomarkers. METHODS: The methods were self-reported smoking compared with salivary thiocyanate and expired carbon monoxide in a 1996, population-based, Swiss survey of 552 men and 565 women. RESULTS: Sensitivity of self-reported smoking relative to salivary thiocynate or carbon monoxide alone was low (38.2% for salivary thiocyanate > or = 100 mg/L, 56.4% for salivary thiocyanate > or = 150 mg/L and 62.6% for carbon monoxide > or = 9 ppm). When defining true positive smokers as people with high concentration of both salivary thiocyanate and carbon monoxide, overall, sensitivity was 88.6% and specificity was 87.2%. In women, sensitivity increased from 85 to 89% when removing subjects exposed to passive smoking. When excluding heavy smokers, sensitivity decreased to 63% in men and to 71% in women. Older women had tendency to misreport smoking. CONCLUSIONS: This comparison of questionnaire data with the simultaneous measurement of salivary thiocyanate and expired carbon monoxide indicates that valid responses can be obtained for self-reported, current smoking in population-based surveys. However, the validity of questionnaires can be underestimated if the gold standard (of exposure to tobacco smoke) is either high levels of carbon monoxide or high levels of salivary thiocyanate.