BACKGROUND AND OBJECTIVE: Uncontrolled bushfires produce copious amounts of smoke that can result in adverse effects on human health and so are important considerations for emergency, public health and environmental protection agencies. During January to March 2003, the north-east and Alpine regions of the state of Victoria experienced major bushfires that created a blanket of smoke over the entire state for extended periods of time. This study aimed to explore the daily trends in air pollutants and temporal correlations with changes in respiratory outcomes and to determine whether variation in particulate matter < 10 microm (PM(10)), visibility-reducing particles and ozone produced by bushfires could explain variation in respiratory outcomes. METHODS: A time-series ecological study was conducted using hospital admission and emergency attendances, air quality and meteorological data. A semi-parametric overdispersed Poisson regression model was used to evaluate the effect of air pollutants. RESULTS: In a semi-parametric analysis of respiratory outcomes adjusted for day-of-the-week and trend effects, only daily levels of PM(10) were strongly associated with emergency department attendances and weakly associated with hospital admissions. CONCLUSIONS: Elevated levels of PM(10) increase the risk for exposed people to attend emergency departments for respiratory conditions. The health effects of bushfire smoke need to be considered when planning health and emergency services and designing public health messages.