Most adult smokers begin smoking during adolescence; nicotine dependence can develop relatively quickly and, once established, most smokers smoke for approximately 40 years. For adolescents dependent upon nicotine, cessation interventions are not well established. It is, therefore, essential that public health interventions focus on preventing initiation and maintenance and decreasing the prevalence of youth smoking. In spite of legislation to protect New Zealand adolescents, a large proportion continues to use tobacco at least weekly. Recent surveys have shown a slight decrease in cigarette smoking prevalence, overall, but, no reduction and marked increases have been reported within some subgroups. The overall aim of this research was to identify school and personal factors associated with secondary school students smoking. The specific research objectives included: a) identifying factors at the personal, family, peer, school and 'tobacco-genic' environment levels which were associated with regular and established cigarette smoking; b) describing the extent of smoke-free policy and health education programmes in secondary schools; and, c) evaluating the relations between cigarette smoking among students and potential protective factors, smoke-free policies and practices and health education programmes. The research was based on data from 3,434 secondary school students from 82 schools. The multi-stage sampling procedures and data analyses ensured that the results were able to be generalised to the New Zealand secondary school student population. Smoking was more prevalent amongst girls for all measures of smoking frequency and significant differences were found for smoking prevalence between ethnic groups and school decile. In terms of family influences, the smoking behaviours of parents were not associated with increased odds of smoking nor were perceived relationships between students and their parents, or exposure to SHS. In contrast, the smoking behaviour of siblings was associated with increased odds of smoking but it is likely that both student and sibling smoking are both influenced by the same processes within the family. Similarly, low levels of self-concept were not associated with increased odds for daily smoking. The smoking behaviour of a best friend was a pervasive risk factor as was a high level of disposable income, frequent episodes of unsupervised activities, and 'pro-smoking' knowledge. Being male, visiting a place of worship, and the intention to stay at school until after Year 13 reduced the odds of daily smoking among students. Multilevel models were used to identify school level effects. After adjusting for student, family and school characteristics significant between-school variance in smoking prevalence remained and this suggests that there are factors, arguably beyond the immediate control of the student or family that may influence a student's smoking behaviour. The presence of a school effect also supports the WHO concept of 'health promoting schools' in that schools can make a difference to health outcomes. Understanding how the health promoting schools model has been interpreted and implemented in NZ schools, along with critique of the implementation of the amended legislation making all schools smoke-free, would be a pertinent 'next-step' in identifying characteristics of schools which are associated with decreased tobacco use.