Abstract The photochemical component of an advanced modelling system for use in pollution control studies is evaluated. Results of a case study for July 1985 in the Lower Fraser Valley region of British Columbia are used to test the performance of the modelling system during an ozone episode. Time series of surface trace pollutant concentrations are used to validate the model prediction. A quantitative analysis of the model errors shows that the modelling system adequately predicts the photochemistry without arbitrary adjustments to the emissions inventory. Improved performance of the ozone forecast would likely be obtained, however, if the non-methane organic compound emissions were increased in urban areas. The mechanistic parameters were shown to be very sensitive to the emissions profile, although the net impact of the parameter changes on ozone production was minor in this case.