Using a statistical approach this paper focuses on an analysis of spatial air temperature variations and their relation to urban land use. The work is based on data collected during an 18-month period at 30 sites in Göteborg, Sweden. Measurements show intra-urban air temperature differences of up to 9 °C in the urban district, and the main purpose of this study is to determine if these variations are statistically significant. A stepwise multiple regression analysis confirmed that surface cover is important for governing air temperature differences in the area. Information on land use and surface cover was gathered from a continuously updated land use database, the Master Plan of Göteborg, and a separate site description analysis. The site description analysis includes a test of three methods using aerial or fish-eye photos for characterisation of surface cover in the urban district. The results show statistically significant temperature variations between different land use/land cover categories on a diurnal basis and for all weather conditions. The importance of the results for urban planning application is discussed.