The study found significant variation in tobacco smoking at both the school and community levels, which indicates the need for a multi-level model in examining factors that influence tobacco smoking. Individual-level predictors were linked to much of the variance at higher levels. The strongest effects stemmed from number of friends who smoke, sibling smoking, and alcohol consumption. Smaller but significant effects were found for smoking by parents. At the community level, socioeconomic disadvantage was significant, but community-level variance in prosocial and drug-related laws/norms was not linked to smoking. The study concludes that cross-level interactions were generally insignificant and that individual risks largely accounted for the substantial variation across schools and communities. Regarding future tobacco-use prevention, findings suggest targeting families in disadvantaged communities. A cross-sectional fixed and random effects model of smoking prevalence was used. The study sample consisted of 7,314 children (53.1 percent female) in grades six and eight who were surveyed in 2006 as part of the Healthy Neighborhoods Study. All measures were based on the Communities That Care youth survey developed in the United States. A two-staged sampling strategy (community and school) was used to select 231 participating schools in 30 communities. 1 table and 46 references