The oxidative potential (OP) of airborne particulate matter (PM) is gaining increasing attention as a health-relevant metric to describe the capacity of PM to promote oxidative stress and cause adverse health effects. To date, most OP studies use filter-based approaches to sample PM and quantify OP, which have relatively poor time resolution (∼ 24 h) and underestimate the contribution of reactive components to OP due to the time delay between sample collection and analysis. To address this important limitation, we have developed a novel instrument which uses a direct-to-reagent sampling approach, providing robust, continuous, high time resolution (5 min) OP quantification, hence overcoming analytical limitations of filter-based techniques. In this study, we deployed this instrument in the Marylebone Road Air Quality Monitoring Station in London, UK, alongside a broad suite of high time resolution PM2.5 composition measurements for three months continuous measurement during Summer 2023. High time resolution OP quantification reveals dynamic changes in volume-normalised (OPv) and mass normalised (OPm) OP evolving over ∼hourly timescales, observed at an average PM2.5 mass concentration of 7.1 ± 4.2 µg m−3, below the WHO interim 4 target of 10 µg m−3. In addition, high time resolution data facilitates directional analysis, allowing us to determine the influence of wind speed and wind direction on OP, and the identification of PM2.5 chemical components and sources which drive dynamic changes in OP; this includes traffic emissions, as well as emissions from the London Underground into the ambient airshed. These results demonstrate the capacity of high time resolution measurements to provide new insights into the temporal evolution of OP, as well as the composition and emission sources which drive OP, developing our understanding of the characteristics of PM2.5 which may promote adverse health impacts.