PurposeThe purpose of this study is to identify correlates of tobacco smoking behaviour across various socio-demographic segments of the Australian population.Design/methodology/approachData from two nationally representative, probability samples of persons 18 and over, surveyed by the Australian Bureau of Statistics in 2001 and 2017–2018 were analysed using multinomial logistic regression.FindingsOverall, the prevalence of current smokers declined from 24.3 to 15%. More than half of the population had never smoked. The prevalence of ex-smokers increased slightly to 30%. Prevalence of current smoking was higher among older age groups and among those with lower educational achievement, lower income, living in a disadvantaged area and experiencing increasing stress. Females were more likely than males to be never-smokers. Males were more likely than females to be current smokers.Research limitations/implicationsThe findings are based on two cross-sectional surveys conducted 17 years apart. It is not possible to draw any conclusions about the actual trajectories of the changes in the values reported or any correlations between those trajectories. Nor is it possible to make any meaningful forecasts about likely future trends in smoking status in these various segments based on these data sets. The classifications used in the surveys generate some heterogeneous groups, which can obscure important differences among respondents within groups. Data are all self-reported, and there is no validation of the self-reported smoking status. This might lead to under-reporting, especially in a community where tobacco smoking is no longer a majority or even a popular habit. Because the surveys are so large, virtually, all the findings are statistically significant. However, the increasing preponderance of never-smokers in many categories might mean that never-smokers could come to dominate the data.Practical implicationsThe findings from this paper will help tobacco-control policy-makers to augment whole-of-community initiatives with individual campaigns designed to be more effective with particular socio-demographic segments. They will also assist in ensuring better alignment between initiatives addressing mental health and tobacco smoking problems facing the community.Originality/valueThe examination of smoking behaviour among individual population sub-groups, chosen by the authors, is commonplace in the literature. This paper uses data from two large surveys to model the whole, heterogeneous population, measured at two different points in time.