Focused on the uncertainty and complexity of risk assessment of karst tunnels and limitation of current methods, the artificial neural network (ANN), wavelet analysis and fuzzy mathematics were integrated. A model of risk assessment of karst tunnels based on fuzzy wavelet neural network (FWNN) was established. According to the advantages and disadvantages of various geophysical methods and the sensitivity of forecasting karst water, and with the method and principle of the comprehensive geological prediction, this study proposed the geological prediction program combined with geological analysis, risk classification, grading prediction and geological disasters approaching alarm technology.