OBJECTIVE:We analysed data collected during a nationwide cross-sectional household survey to estimate the prevalence of and identify factors associated with smoking in Pakistan. DESIGN:Population-based, cross-sectional survey [National Health Survey of Pakistan (NHSP) 1990-1994]. METHODS:A population-based survey was carried out in Pakistan during 1990-1994. A nationally representative sample of 18,135 individuals aged 6 months and older was surveyed. We restricted this analysis to individuals aged 15 years or older (n=9442). The main outcome measure was self-reported smoking. Smokers were defined as individuals who reported current smoking and having smoked at least 100 cigarettes or 'beddies' during their lifetime. RESULTS:Overall prevalence of smoking was 15.2% [95% confidence interval (CI), 14.5-15.9%]. It was 28.6% (27.3-29.9%) among men and 3.4% (2.9-3.9%) among women. The highest prevalence was reported in men aged 40-49 years (40.9%). The independent predictors of smoking identified in the multivariate logistic regression analysis included age, male gender, ethnicity and illiteracy. CONCLUSIONS:One out of every two to three middle-aged men in Pakistan smoke cigarettes. Our findings suggest that ethnically sensitive smoking control programmes that include measures for improving literacy rates are needed in Pakistan.