Under the scenario of global warming,the air temperature has changed significant in the mid-high latitude regions of North Hemisphere. The climate in Mongolia plays an important role in the East Asia climate system. However,the unavailability of meteorological observations in Mongolia handicapped the understanding on the climate changes in this country. Reanalysis products,surrogates for large-scale observations,have been increasingly applied to assess climate impact in the past two decades because they are designed to estimate the state of real atmosphere and land surface characteristics. In order to evaluate the applicability of the JRA-55 and the ERA-interim products of air temperature in Mongolia,this study compared them with the observations of 18 metrological stations in Mongolia from January 1,1987 to December 31,2006. High correlations with an average of 0. 99 indicate that the two datasets could capture the annual cycle very well. However,the average root-meansquare error(RMSE)of 2. 5°C and 2. 3°C for JRA-55 and ERA-interim respectively,reveals that they could not be applied directly for the individual sites. The biases can be mainly attributed to the altitude differences between ERA-Interim grid points and stations. An elevation correction method based on monthly lapse rates is limited to reduce the bias for all stations. After the calibration,the biases in all seasons except the winter were largely reduced. Furthermore,the seasonal and annual trends of the reanalysis data of air temperature were analyzed and compared with those of the observations. The results indicated that the two kinds of reanalysis monthly air temperature data in Mongolia agree well in the quality. The biases of ERA-interim are smaller than JRA-55,while JRA-55 can effectively capture the variability of air temperature.