Smoking is a major health hazard. Most cigarette smokers start by the age of 18 years. The purpose of this study was to assess the prevalence of the intention to smoke among the students of a metropolitan compared to a non-metropolitan high school. The influence of age, sex, demographic and socio-economic variables, and the role of smoking models of family members and friends, were examined. Nine hundred forty-five students (529 males and 416 females; mean age 15.8 +/- 1.5 years) attending a high school in Naples and 442 students (223 males and 219 females; mean age 16.1 +/- 1.6 years) in Capua, a small town 40 Km distant from Naples, filled in an extensive questionnaire on smoking. The prevalence of intention to smoke was 10.4% in Naples and 9.3% in Capua. It was related to age (p < 0.01) in Naples, but not in Capua. The prevalence of smokers was 24.2% in Naples (males 21.6%, females 27.6%; p = 0.038) and 24.1% in Capua (males 29.2%, females 19%; p = 0.017). As expected, in both cities intention to smoke was associated (p < 0.001) with the strength of existing smoking habit. Students smoking over 21 cigarettes/week were more likely to continue than students smoking less 21 cigarettes/week, both in Naples and in Capua. More than half of smoking students, in both cities, were irresolute about their habit in the subsequent year. In Naples, intention to smoke of male students was associated with mother's (p = 0.02) and siblings' (p < 0.0001) smoking habit; in female students intention to smoke was associated with father's (p = 0.02), mother's (p < 0.001), parents' (p < 0.01) and siblings' smoking habit (p = 0.0002). In Capua an association was evident, in male students, between intention to smoke and paternal smoking habit (p = 0.04); in female students, intention to smoke was associated with siblings' smoking habit (p = 0.03). In Naples and in Capua, for both sexes, intention to smoke was related to smoking habits of the best friend of the same sex (p < 0.0005), the best friend of the opposite sex (p < 0.00005) and friends (p < 0.00001). Multivariate analysis showed, in Naples, an independent relation between adolescent intention to smoke and age (p = 0.01), smoking status of student (p < 0.0001) and friends' smoking habit (p = 0.01). In male students intention to smoke was associated with age (p = 0.003), smoking habit of student (p < 0.0001), mother's (p = 0.02) and friends' (p = 0.02), whereas in females it was associated with smoking behavior of student (p < 0.0001). In Capua student intention to smoke was related to the smoking status of the student (p < 0.0001) and of the best friend of the opposite sex (p < 0.04); in male as in female students, intention to smoke was associated with smoking habit of the student (p < 0.0001). In conclusion, prevalence of adolescents' intention to smoke is similar in two distinct populations of high school students of a city and a small town. Smoking is at higher prevalence among females in the city and among males in the small town. Intention to smoke increases with age, in the great city, and is related to student's existing habit and peer models. More than half of smoking students, in both cities, were irresolute about their habit in the subsequent year. This study has identified some variables associated with adolescents' intention to smoke; we feel that these findings may contribute to a better understanding of smoking behavior among adolescents and may have preventive implications.