An important tool in the management of photochemical smog episodes is a computational model which can be used to test the effect of possible emission control strategies. High spatial resolution of such a model is important to reduce the impact of numerical errors on predictions and to allow better comparison of the model with experimental data during validation. This paper therefore presents the development of an adaptive grid model for the Central European Region describing the formation of photochemical oxidants based on unstructured grids. Using adaptive methods, grid resolutions of less than 20 km can be achieved in a computationally effective way. Initial simulation of the photochemical episode of August 1998 indicates that the model captures the spatial and temporal tendencies of ozone production and demonstrates the effectiveness of adaptive methods for achieving high resolution model predictions.