Abstract: A new warning scheme of solar proton events was proposed in this paper. In this scheme, five parameters including the solar flare position in the sun, the peak flux of X ray radiations and its rise bine, the half integrated fluxes of microwave fadiations of 2695MHz and 8800MHz, were used to give a warning of the occurences of solar proton events and their peak flux and time with an ANN (Artificial Neural Network) method. A simple discription of ANN was given. Some historical data of solar flares and solar proton events were provided for ANN's training and test repectively. Results showed that the presented scheme has a prediction accuracy of 93.75%.