Studying the changes in hydrological conditions in the middle reaches of the Xijiang River can provide hydrological data support for high-quality development of the economy and society，rational utilization of water resources，and construction of water conservancy projects in the Xijiang River Basin.The Mann-Kendall trend analysis method，Mann-Whitney-Pettitt change-point detection method，and Morlet wavelet analysis method were employed to conduct the analysis of trend，change-point，and periodicity，respectively；Meanwhile，the predictions of annual average water level and annual runoff were made based on BP neural network.The results showed that the annual average temperature displayed a significant increasing trend，the annual average water level showed a significant decreasing trend，and the changes in other hydrological factors were not obvious；both the annual average water level in 2002 and the annual average temperature in 2014 experienced significant changes，while other hydrological factors did not show significant changes；the BP neural network models for annual average water level and annual runoff in the middle reaches of the Xijiang River showed good application effects.