Abstract Background Severe air pollution generated by forest fires is becoming an increasingly frequent public health management problem. We measured the association between forest fire smoke events and hospital emergency department (ED) attendances in Sydney from 1996–2007. Methods A smoke event occurred when forest fires caused the daily citywide average concentration of particulate matter (PM 10 or PM 2.5 ) to exceed the 99th percentile of the entire study period. We used a time-stratified case-crossover design and conditional logistic regression models adjusted for meteorology, influenza epidemics, and holidays to estimate odds ratios (OR) and 95% confidence intervals (CI) for ED attendances on event days compared with non-event days for all non-trauma ED attendances and selected cardiorespiratory conditions. Results The 46 validated fire smoke event days during the study period were associated with same day increases in ED attendances for all non-trauma conditions (1.03, 95% CI 1.02, 1.04), respiratory conditions (OR 1.07, 95% CI 1.04, 1.10), asthma (OR 1.23, 95% CI 1.15, 1.30), and chronic obstructive pulmonary disease (OR 1.12, 95% CI 1.02, 1.24). Positive associations persisted for one to three days after the event. Ischaemic heart disease ED attendances were increased at a lag of two days (OR 1.07, 95% CI 1.01, 1.15) while arrhythmias had an inverse association at a lag of two days (OR 0.91, 95% CI 0.83, 0.99). In age-specific analyses, no associations present in children less than 15 years of age for any outcome, although a non-significant trend towards a positive association was seen with childhood asthma. A further association between smoke event and heart failure attendances was present for the 15–65 year age group, but not older adults at a lag of two days (OR 1.37 95% CI 1.05, 1.78). Conclusion Smoke events were associated with an immediate increase in presentations for respiratory conditions and a lagged increase in attendances for ischaemic heart disease and heart failure. Respiratory impacts were either absent or considerably attenuated in those