Wildfire smoke can have significant health effects, in the short and long term, with some estimates suggesting that smoke may affect the health of 1 in 3 residents in the United States. This highlights the importance of being able to accurately model the extent and severity of particulates resulting from wildfire smoke to inform during-fire decisions and a smoke management planning for prescribed fires. Forest Service scientists tested the idea that social media data can be a complementary approach to modeling air pollution caused by wildfires. Conversations about smoke and wildfire on Twitter were used to model air quality impacts of the King Fire, which occurred in northern California in the fall of 2014. They found that users’ tweets could accurately predict air quality data collected by physical monitoring stations in the region, even when controlling for geographic and temporal variables. Analyses of these conversations also revealed people’s reactions and beliefs pertaining to fires; for example, people closest to the fire were most concerned about the health impacts of the smoke. This project provides evidence that computational social science approaches could provide an alternative way of assessing air quality, particularly in remote regions where physical monitoring stations may not be present.