Water prediction is the basis for water resource planning and management. However, water resource systems are complex. Water consumption is influenced by various factors whose relations are also complicated. The degree of influence is always different for the same factor in different areas. The effective factors of water consumption are analyzed thoroughly. The influencing factors of high degree are selected to establish an improved coupling model of grey system and multiple regressions to predict water consumption in Wuhan. The coupling model is clear in concept, simple in structure, and convenient in use. The complex relationship between water consumption and its main influencing factors is reflected. The model has the potential advantage for predicting annual water consumption. The applied research in Wuhan showed that the forecast effect of improved coupled model is good with relative error less than 1%. The model is used to predict water consumption of 2015 in Wuhan as 4.1430424 billion tons.