OBJECTIVE. This study examined the change of smoking behavior among Thai youth and the relationship between individual factors, surrounding factors and health behavior factors, and smoking behavior of the youth in the Kanchanaburi Demographic Surveillance System area (KDSS) during the years 2003-2004. METHODS. This study used secondary data from the KDSS survey and the sample for this study included a total of 2,510 youth aged 15-24 years. Pooling crosssectional data for these two years across time, the number of cases became totally 5,020 cases. The reason for using independently pooled cross-section data is to increase the sample size at different points in time so we can get more precise estimators or test statistical significance with more power. Observations within clusters more likely had similar characteristics or intracluster correlations embedded in the data structure. So robust standard errors estimate for cluster sampling data in logistic modeling had to be taken into account. RESULTS. The total smoking rate increased slightly from 13% in the year 2003 to 16% in the year 2004. There was a marked increase in the smoking rate among male youth from 25% in the year 2003 to 33% in the year 2004. Smoking rates were very high among those 20-24 years of age, low-educated, from poor households, upland stratum, non-agricultural occupations, those whose father smoked, households with at least one other smoker, those from villages with a moderate and high percentage of people smoking, and among those who also drank alcohol. Logistic regression analysis also pointed out that the most statically significant factors related to youth smoking in the Kanchanaburi DSS were sex, age at first starting to regularly smoke, education, marital status, household economic status, occupation, father smoking, percentage of other people who smoke in the village, number of household members smoking, number of shops in the village, drinking alcohol and time. CONCLUSIONS. This study pointed out that most youth smokers had different characteristics when compared with youth who did not smoke, such as individual factors including sex, education, marital status, age at first starting to smoke, and household economic status; and environmental factors including a smoking father, number of smoking household members, and percent of smokers in the community. Moreover, the study showed that youth who drank alcohol were much more likely to smoke than youth who did not drink alcohol. Therefore, campaigns to stop smoking and drinking alcohol should be combined for all ages and have to focus on parents, family and community members including youth/ peers from the time that they are young and educating them about the harm of smoking and its impacts on their health.