The invention provides a water quality prediction method and system. According to the method, an ARIMA auto-regression integration sliding average model is utilized in combination with a BP neural network to predict water quality time sequence data. The method is advantaged in that plenty of water quality data of a to-be-predicted water area can be predicted, large prediction scope, high precision and fast speed are realized, and multi-water source monitoring, water quality early warning and water pollution treatment are facilitated.