Variations in Asian upper tropospheric temperature during summer is closely related to, and may, indeed, serve as a useful predictor of East Asian precipitation. The predictability of these interannual variations in summer UTT (upper tropospheric temperature, represented by 500-200 hPa thickness) for the period 1960-2005 in the ENSEMBLES multi-model seasonal forecast, initiated 1 May every year, was examined in this study. Results showed that the interannual variability of Asian UTT in summer was skillfully predicted by ENSEMBLES, as measured by the good prediction of its standard deviation centers in the mid-latitude and high correlation coefficient of its first two leading interannual variability modes compared with observations. The main deficiency of the multimodel ensemble mean (MME) was that the temperature at high-latitudes could not be captured. The correlation coefficients of the first (PC1) and second (PC2) principle components in the MME with those from NCEP/NCAR (National Centers for Environmental Prediction/National Center for Atmospheric Research) reanalysis were 0.63 and 0.77, respectively. The first two leading models of Asian UTT in summer in observation were dominated by (1) the silk-road teleconnection at the upper troposphere forced by Indian monsoon precipitation anomalies in an ENSO-developing summer, and (2) the Pacific-Japan teleconnection forced by northwestern Pacific Ocean rainfall anomalies in an ENSO-decaying summer, respectively. These processes were well-predicted by ENSEMBLES; thus, a high prediction skill for Asian summer UTT was shown in ENSEMBLES. The first two leading modes of Asian summer UTT well-represented the zonal and meridional thermal contract variation. A comparison with two previous widely-used East Asian Summer Monsoon (EASM) indices was performed; results showed a much better predictive outcome for PC1 than for the traditional zonal thermal contrast. Using PC1 of Asian summer UTT and the traditional meridional EASM index as two predictors of summer precipitation over eastern China, a dynamical-statistical forecast model was established. The cross-validation results showed that the new forecast model significantly improved the predictive skill of summer precipitation over Northeast China and the upper stream of the Yangtze River. The first leading mode of Asian summer UTT and corresponding PC could well represent the zonal thermal contrast, which has a correlative relationship with summer precipitation over eastern China, and can be well-predicted by climate models. It can effectively serve as one predictor of summer precipitation over mid-latitude China, particularly northeastern China.