In this study a new method using Artificial Neural Networks has been used to predict design parameters of grouting. In this method, a single layer perceptron, one of the most practical kinds of artificial networks, is used. For learning and test of network, data of Ardabil dam is used and by using mean - standard deviation and linear transfer methods, data has been normalized. Using assessment criteria, the optimum number of neurons for middle layers is chosen. Depth parameter, RQD, water pressure and lugeon for input parameters, and grouting duration, grouting pressure and cement usage for output parameters are applied into the network. Although this prediction is available for new holes in the same dam, it can be applied to other dams with different foundation properties.