A three-dimensional, non-hydrostatic, high-resolution numerical model was used to analyse urban heat-island (UHI) intensity in an idealised but realistic configuration. The urban area was 20 km square and lay on flat land at about latitude 50degrees N in a maritime climate. In the model the urban area was represented by anomalies of albedo, anthropogenic heat flux, emissivity, roughness length, sky-view factor (SVF), surface resistance to evaporation (SRE) and thermal inertia. A control simulation included all these factors and the resultant UHI structure, energetics and intensity were validated against observations. The results also compared favourably with earlier simulations. A series of experiments was conducted in which successively one of the anomalies that represented the urban area was omitted from the control simulation so as to provide the basis for an assessment of its effect. In daytime the individual effects due to albedo, anthropogenic heat, emissivity, SVF and thermal inertia ranged from 0.2 to 0.8 degreesC. In common with albedo, anthropogenic heat, emissivity and SVF, the SRE aided the formation of a UHI; it was also the most important factor in increasing its intensity. The roughness length had the opposite effect. At night emissivity, roughness length, SVF and SRE had effects ranging from 0.3 to 0.75 degreesC, but the largest effect (2 degreesC) was due to the anthropogenic heat. These results showed a difference in the causes of daytime and nighttime UHIs. In daytime the roughness length and SRE were the most important factors affecting UHI intensity; at night the anthropogenic heat was the most important. The simulations suggested that the size of the urban area had a minimal effect on UHI intensity.