Biogenic volatile organic compounds (BVOCs) influence atmospheric composition and climate, and their emissions are affected by changes in land use and land cover (LULC). Current Earth system models calculate BVOC emissions using parameterisations involving surface temperature, photosynthetic activity, CO2 and vegetation type and use emission factors (EFs) to represent the influence of vegetation on BVOC emissions. We present new EFs for the Interactive BVOC Emission Scheme (iBVOC) used in the United Kingdom Earth System Model (UKESM), based on those used by the Model of Emissions of Gases and Aerosols from Nature (MEGAN) v2.1 scheme. Our new EFs provide an alternative to the current EFs used in iBVOC, which are derived from older versions of MEGAN and the Organizing Carbon and Hydrology in Dynamic Ecosystem (ORCHIDEE) emission scheme. We show that current EFs used by iBVOC result in an overestimation of isoprene emissions from grasses, particularly C4 grasses, due to an oversimplification that incorporates the EF of shrubs (high isoprene emitters) into the EF for C3 and C4 grasses (low isoprene emitters). The current approach in iBVOCs assumes that C4 grasses are responsible for 40 % of total simulated isoprene emissions in the present day, which is much higher than other estimates of ∼ 0.3 %–10 %. Our new isoprene EFs substantially reduce the amount of isoprene emitted by C4 grasslands, in line with observational studies and other modelling approaches, while also improving the emissions from other known sources, such as tropical broadleaf trees. Similar results are found from the change to the terpene EF. With the new EFs, total global isoprene and terpene emissions are within the range suggested by the literature. While the existing model biases in the isoprene column are slightly exacerbated with the new EFs, other drivers of this bias are also noted. The disaggregation of shrub and grass EFs provides a more faithful description of the contribution of different vegetation types to BVOC emissions, which is critical for understanding BVOC emissions in the pre-industrial and under different future LULC scenarios, such as those involving wide-scale reforestation or deforestation. Our work highlights the importance of using updated and accurate EFs to improve the representation of BVOC emissions in Earth system models and provides a foundation for further improvements in this area.