Chronic exposure to fine particulate matter (PM2.5) poses a major global health risk, commonly assessed by assuming equivalent toxicity for different PM2.5 constituents. We used a data-informed global atmospheric model and recent exposure-response functions to calculate the health burden of ambient PM2.5 from ten source categories. We estimate 4.23 (95% confidence interval 3.0-6.14) million excess deaths annually from the exposure to ambient PM2.5. We distinguished contributions and major sources of black carbon (BC), primary organic aerosols (POA) and anthropogenic secondary organic aerosols (aSOA). These components make up to ∼20% of the total PM2.5 in South and East Asia and East Africa. We find that domestic energy use by the burning of solid biofuels is the largest contributor to ambient BC, POA and aSOA globally. Epidemiological and toxicological studies indicate that these compounds may be relatively more hazardous than other PM2.5 compounds such as soluble salts, related to their high potential to inflict oxidative stress. We performed sensitivity analyses by considering these species to be more harmful compared to other compounds in PM2.5, as suggested by their oxidative potential using a range of potential relative risks. These analyses show that domestic energy use emerges as the leading cause of excess mortality attributable to ambient PM2.5, notably in Asia and Africa. We acknowledge the uncertainties inherent in our assumed enhanced toxicity of the anthropogenic organic and BC aerosol components, which suggest the need to better understand the mechanisms and magnitude of the associated health risks and the consequences for regulatory policies. However our assessment of the importance of emissions from domestic energy use as a cause of premature mortality is robust to a range of assumptions about the magnitude of the excess risk.