Objectives: To determine the impact of bushfires on children’s physical activity.

Design: Natural experiment comparing device-measured physical activity and air quality index data for schools exposed and not exposed to the Australian bushfires.

Methods: Participants were drawn from 22 schools participating in a cluster randomised controlled trial of a school-based physical activity intervention that coincided with the 2019 Australian bushfires. Students in Years 3 and 4 (8–10 years old) provided data. We used propensity score matching to match 245 exposed and 344 control participants.

Main outcome measures: Minutes of moderate and vigorous physical activity.

Results: The bushfires had minimal effect on children’s average weekly physical activity. Analysis of acute effects showed children maintained their levels of physical activity up to an estimated turning point of air quality index of 737.08 (95% CI = 638.63, 835.53), beyond which daily physical activity levels dropped sharply. Similar results were found for girls and boys and for children from low-to-average and higher socio-economic backgrounds.

Conclusions: Children’s physical activity was not strongly influenced by the presence of smoke and targeted public health advice during the bushfires might not have had the intended effect of reducing children’s outdoor physical activity. Only when air quality deteriorated to approximately 3.5 times the Air Quality index threshold (>200) deemed ‘hazardous’ by the Australian Department of Health did children’s physical activity decline. Public health agencies should re-evaluate the effectiveness of health messages during bushfires and develop strategies to mitigate risks to children’s health.