Background Estimates of road vehicle exhaust emissions are some of the key inputs to air pollution dispersion models which are used to estimate ambient air pollution levels, human exposure to traffic-related air pollution, and associated health effects and impacts. Current UK vehicle emission estimates for hot exhaust pollutants are derived using functions of the average vehicle speed, over a trip. These average-speed-emission functions are often sourced from a widely used regional European emission model referred to as COPERT. COPERT has been repeatedly shown to underestimate actual exhaust emissions that occur in real-world driving. We aimed to produce a set of more reliable and transparent average-speed-emission functions for all vehicle types and EURO emission standards operating in Bradford, UK. Methods We undertook thirty-three hours of real-world driving over Bradford, UK, in an instrumented Toyota Prius (Euro 5, 2.0l petrol-electric hybrid) equipped with a VBoxII Lite and an OBD Mini data logger; both of which logged the vehicle instantaneous speeds over tracked journeys. The collected instantaneous speeds were verified and fed into the Instantaneous Emission Model PHEM which estimated the second-by-second emissions of NO x for 167 vehicle types and emissions standards. Using these outputs, and following a novel micro-trip approach which splits the driving cycles into smaller driving events contained between adjacent stationary periods (stop-start), new vehicle average-speed-emissions functions were developed. These functions were then linked to speeds and flows modelled and validated in Bradford and were used to estimate NO x road transport emission inventory. Estimates were also derived from COPERT. Results We developed 167 new NO x average-speed-emission functions for the full fleet in Bardford. Compared to COPERT, our functions produced higher emission estimates at the lower speeds which incorporate significant proportions of stop-start driving. Our functions estimated that in the AM peak petrol cars (55.43% of all vehicles) produced 12% of all NO x emissions, diesel cars (26.83% of all vehicles) produced 25%, diesel buses and coaches (1.55% of all vehicles) produced 21%, light duty vehicles (12.54%) produced 14% and heavy duty vehicles (2.13% of all vehicles) produced 27% of NO x . Conclusions This work has shown important differences between predicted emissions from both models and suggested that COPERT may significantly under-predict the air pollution contribution from slower traffic. More data is needed at the higher speed segments and both models need further validation. The two models predicted different emission contributions from the different vehicle classes and this has consequences for health improvement policies.