OBJECTIVES In the etiology of childhood cancers, many genetic and environmental factors play a role. One of these factors is cigarette smoking and the main source of tobacco smoke exposure of children is parental smoking. However, establishing a causal relationship between parental smoking and childhood cancers has proven challenging due to difficulties in accurately detecting tobacco smoke exposure METHODS To address this issue, we used hair cotinine analysis and a questionnaire to get information about tobacco smoke exposures of pediatric cancer patients and healthy children. 104 pediatric cancer patients and 99 healthy children participated in our study. Parental smoking behaviours (pre-conceptional, during pregnancy and current smoking) and environmental tobacco smoke (ETS) exposures of children are compared. RESULTS We have found no differences between two groups by means of maternal smoking behaviours. However, the rates of paternal pre-conceptional smoking and smoking during pregnancy were significantly low in cancer patients (p<0,05). These data suggests that social desirability bias among fathers of cancer patients may have contributed to this discrepancy. According to questionnaire cancer patients had significantly lower ETS exposures than healthy children (p<0,05). However, ETS exposure assessment through cotinine analysis demonstrated that cancer patients had higher exposure to ETS compared to healthy children (p<0.001). CONCLUSION Our findings provide evidence supporting the potential role of smoking as a risk factor for childhood cancers. This study also revealed that questionnaires could cause biases. We suggest that cotinine analysis along with validated questionnaires can be used to prevent biases in studies of tobacco smoke in the etiology of childhood cancers.