Accurate identification of smoking behaviour is crucial to monitor the smoking rate. This study used urinary cotinine (UC) as a biomarker to verify the effectiveness of self-reported smoking behaviour among conscripts during recruit training. The influence of second-hand smoke (SHS) on the UC concentration was also analysed. A cross-sectional study was conducted from July 2014 to December 2014. The participants comprised a total of 621 military service and basic military training conscripts. A self-administered questionnaire survey and a urine test were performed to verify the participants' smoking behaviour. The UC concentration of 100 ng/mL was adopted as the baseline to identify smokers. A high level of consistency was observed between the conscripts' self-reported results and the results validated by the UC concentrations (the overall kappa coefficient was 0.918). Moreover, the overall sensitivity and specificity were 92.9% and 98.1%, respectively. The sensitivity for the military service conscripts was significantly lower than that for the basic military training conscripts (86.1% vs. 97.5%, P-value = 0.002). For the self-reported nonsmokers among the military service conscripts, SHS exposure was related to their UC concentrations. The method of self-reporting through a questionnaire survey can serve as a tool to assess conscripts' smoking behaviour.