In view of the load density selection problem in spatial load forecasting in the distribution network， a load density index selection method based on the interval analytic hierarchy process（IAHP） and technique for order preference by similarity to ideal solution（TOPSIS） methods is proposed. The method is an improvement on the traditional analytic hierarchy process for using interval data instead of dot data while effectively combining the theory of the IAHP method with the TOPSIS method， and the advantages of human expertise with quantitative calculation in rationally handling the uncertainties of decision factors and fuzziness of experts judgment. This has enhanced the reliability of prediction results of load density. Finally， the method of similarity is used to correct the load density， further improving the precision of forecast. A case study shows that the method proposed has better prediction results and smaller errors.