Objective: The purpose of this cross-sectional study was to determine status of smoking among youths in school and to identify both student-level and school-level factors on smoking.
Material and Method: A stratified two stage cluster sampling was employed to recruit 2,557 students in grades 10-12 and 30 teachers from 30 high schools in academic years 2010 in Buriram province, using a multilevel logistic analysis.
Results: The current smoking status was 9.0% overall, 19.0% and 1.3% for male and female respectively. The results from multilevel logistic analysis revealed that only 10.5% of variation in smoking was associated with difference in school characteristics. The following student-level factors entered into the multilevel logistic regression model according to their importance were self-efficacy-with the moderate and low self-efficacy having a higher risk of smoking compared with high (OR 8.76, 5.80)-alcohol drinking (OR 5.42). Males were more likely to smoke than females (OR 4.56), perceived benefit on non-smoking-with the moderate and low perception having a greater chance of smoking compared with the higher level of perception (OR 2.31, 1.93). School-level factors including type of school, health promoting school policy, student-teacher ratio, proportion of teachers who smoked and availability of cigarettes near the school were also taken into account, but were not related to the status of smoking among the students.
Conclusion: The present study confirmed that school factors had no significant relationship to smoking among youths. Strict smoke-free policy at school is recommended. School program should focus more on self-efficacy. Social marketing campaign/education should focus on friends and family as non-smoking role models.

Keywords: Multilevel, Smoking, Youths in school