The heat island phenomenon is surveyed. Existing theories are criticized as being excessively qualitative. A general purpose, numerical energy budget model is described and applied to the urban atmosphere. Calculations for several special cases as well as a sensitivity analysis are presented. The model is found to predict the correct order of magnitude of the urban temperature excess. The heat island effect is found to be the net result of several competing physical processes. In general, reduced evaporation in the city center and the thermal properties of the city building and paving materials are the dominant parameters. It is suggested that such a model could be used in engineering calculations to improve the climate of existing and future cities.