Based on the measured data in 101 meteorological stations over the Yellow River Basin, and the precipitation and temperature data dating from 1961 to 2099 of the six CMIP5 models under three scenarias, statistical downscale study is conducted by using equidistant cumulative distribution function matching method (EDCDFm). The accuracy of the downscaled data is evaluated by validating against the measured data dating from 1961 to 2005. After that, the meteorological drought around the Yellow River Basin is predicted by means of standard precipitation index (SPI). The results show that the accuracy of the raw models simulated temperature and precipitation has been significantly improved by using EDCDFm method, it is especially true with the modelling accuracy of their extreme values. The predicted results of the drought around the Yellow River Basin indicate that, the drought condition at the beginning of the 21st century has become more severe than that of the reference period under three scenarios, but is likely to be obviously mitigated at the end of this century. Notwithstanding, preventing drought around the Yellow River Basin is still a tough job in recent years.