An optimization model used to determine the optimum insulation thickness (OIT) of walls and roofs of existing buildings was established. Primary energy saving ratio (PESR), global cost saving ratio (GCSR) and pollutant emission reduction ratio (PERK) were used as evaluation criteria. Least square method was used to fit the polynomials which indicated the relationship between annual air-conditioning demand and insulation thickness. An existing building was used as the basis of the case study. The average relative error on annual heating and cooling demand was 0.23% and 0.035%, respectively, which indicated the accuracy of both fitting polynomials. Four types of heat and cold sources were considered to be used for the case building, respectively. Sensitivity analysis was carried out to investigate the impact of economic and energy factors on the optimization results. Results showed that the OIT of walls and roofs could be obtained by using the optimization model. The OIT of walls and roofs, building performance indicators and overall evaluation criterion during building envelope retrofit depended on the weight coefficients of evaluation criteria and types of heat and cold sources. The impact of sensitivity factors on the optimization results for the four types of heat and cold sources were different. (C) 2018 Elsevier Ltd. All rights reserved.