Particulate matter (PM) has been considered to be closely related to human health, especially fine particulate matter. However, whether PM mass concentration alone is a good indicator for health impact remains a challenging question. In this study, emissions from residential coal combustion (RCC), one of the important PM sources in northern China, were tested to examine the relationship between the emission factors of particle-generated reactive oxygen species (ROS) (EF ROS) and PM (EF PM). A total of 24 combinations of source tests were conducted, including eight types of coal with different geological maturities (two anthracites and six bituminous) burned in three types of stoves (one honeycomb coal stove, one old chunk stove, and one new chunk stove). Here, ROS was defined as generated hydroxyl radical (·OH) by PM, and results showed EF ROS from 24 residential coal combustion varied greatly by nearly 20 times. EF ROS ranged 0.78–14.85 and 2.99–12.91 mg kg−1 for the emissions from honeycomb and chunk coals, respectively. Moreover, the correlation between EF ROS and EF PM was significantly positive in honeycomb coal emissions (r = 0.82, p < 0.05), but it was insignificant in chunk coal emissions (r = 0.07, p > 0.05). For honeycomb coal emissions, organic carbon (OC) was quite abundant in PM and it might be the predominant contributor to both EF PM and EF ROS , resulting in a strong and positive correlation. For chunk coal emissions, high EF ROS was mainly related to relatively high metal emissions in AN and LVB, while the metals were not major components in PM, leading to a poor correlation between EF PM and EF ROS. Therefore, this study revealed that PM was not always positively correlated with ROS from residential coal burning, and the relationship was mainly determined by the compositions of PM, suggesting PM mass concentration alone may not be the best indicator for assessing health impacts. • Emission factors of ROS may vary by 20 times in different domestic coal burning. • EF ROS and EF PM did not always correlate, depending on the composition of PM. • PM mass concentration may be not a good indicator for health impact. [ABSTRACT FROM AUTHOR]