To determine the air-conditioning outdoor design temperature,the generalized Pareto distribution(GPD) was used to study the probability distribution of outdoor air extremely high wet- and dry-bulb temperatures observation data.Firstly,taking Tianjin as an example,based on the 30 years climate data,the GPD model of extremely high temperatures was established and the air-conditioning outdoor design temperatures were calculated with 50 h non-guaranteed.And then,the applicability of GPD method was verified from three perspectives:data acquisition frequency,climate change and climate types.The results indicate that GPD density curve has a good agreement with empirical distribution curve in Tianjin.The difference in outdoor air-conditioning design temperatures detween GPD method and standard method is no more than 0.1,℃.GPD method is insensitive to data acquisition frequency. Besides,the design temperatures on the bases of hourly data and timing data are calculated by GPD method,the biggest difference being 0.1,℃ for dry-bulb temperature.GPD method can reflect the influence of climate change and is suitable for describing the probability distribution of meteorological parameters with different climate features.In summary,GPD method can be applied to statistical determination of summer outdoor design temperatures of airconditioning.