Local air quality management requires the use of screening and advanced modelling tools that are able to predict roadside pollution levels under a variety of meteorological and traffic conditions. So far, more than 200 air pollution hotspots have been identified by local authorities in the UK, many of them associated with NO2 and/or PM10 exceedences in heavily trafficked urban streets that may be classified as street canyons or canyon intersections. This is due to the increased traffic-related emissions and reduced natural ventilation in such streets. Specialised dispersion models and empirical adjustment factors have been commonly used to account for the entrapment of pollutants in street canyons. However, most of the available operational tools have been validated using experimental datasets from relatively deep canyons (H/W≥1) from continental Europe. The particular characteristics of low-rise street canyons (H/W<1), which are a typical feature of urban/sub-urban areas in the UK, have been rarely taken into account. The main objective of this study is to review current practice and evaluate three widely used regulatory dispersion models, WinOSPM, ADMS-Urban 2.0 and AEOLIUS Full. The model evaluation relied on two comprehensive datasets, which included CO, PM10 and NOx measurements, traffic information and relevant meteorological data from two busy street canyons in Birmingham and London for a 1-year period. The performance of the selected models was tested for different times of the day/days of the week and varying wind conditions. Furthermore, the ability of the models to reproduce roadside NO2/NOx concentration ratios using simplified chemistry schemes was evaluated for one of the sites. Finally, advantages and limitations of the current regulatory street canyon modelling practice in the UK, as well as needs for future research, have been identified and discussed. © 2007 Elsevier Ltd. All rights reserved.