Storm and flood such as torrential rains and major typhoons has often caused damages on a large scale in Korea and damages from storm and flood have been increasing by climate change and warming. Therefore, it is an essential work to maneuver preemptively against risks and damages from storm and flood by predicting the possibility and scale of the disaster. Generally the research on numerical model based on statistical methods for analyzing and predicting disaster risks and damages has been mainstreamed. In this paper, we developed the model for prediction of damage cost from storm and flood by the neural network algorithm which outstandingly implements the pattern recognition. Using the damage data of storm and flood and meteorological data from 1991 to 2005 in Korea, we made data sets and defined the accumulated rainfall, the maximum wind speed, the antecedent rainfall within 5 days before being disasters, and the regional feature representing the influence factors on the outbreak of damages from storm and flood as input variables for learning the model. Also we defined the total amount of damages as an output variable. Creation of a holdout which was created by randomly partitioning into train, validation, and test data in the ratio of 6:3:1 respectively was repeatedly processed by 5 times. For finding the optimal model, first of all, we learned about 8,000 models initialized by combinations of the architectures: initial weight and the number of nodes in a hidden layer, and learning parameters: momentum and learning rate of a neural network model. Next, we selected the candidate models for an optimal model among the learned models according to the accuracy and TPR and FPR on ROC graph. Finally, we decided the optimal model for predicting damage cost from storm and flood among the candidate models according to the accuracy and TPR and FPR on ROC graph obtained using test data.