Correct estimation of the scour around vertical piles in the field exposed to oscillatory waves is very important for many offshore structures and coastal engineering projects. Conventional predictive formulas for the geometric properties of scour hole, however, are not able to provide sufficiently accurate results. Artificial Neural Networks (ANNs) are simplified mathematical representation of the human brain. Three-layer normal feed-forward ANN is a powerful tool for input-output mapping and has been widely used in civil engineering problems. In this article the ANNs approach is used to predict the geometric properties of the scour around vertical pile. Two different ANNs including multilayer perceptron (with four different learning rules) and radial basis functions neural networks are used for this purpose. The results show that a three-layer normal feed-forward multilayer perceptron with quick propagation (QP) learning rule can predict the scour hole properties successfully. Correct estimation of the scour around vertical piles in the field exposed to oscillatory waves is very important for many offshore structures and coastal engineering projects. Conventional predictive formulas for the geometric properties of scour hole, however, are not able to provide adequate accurate results. Artificial Neural Networks (ANNs) are simplified mathematical representation of the human brain. Three-layer normal feed-forward ANN is a powerful tool for input-output mapping and has been widely used in civil engineering problems. In this article the ANNs approach is used to predict the geometric properties of the scour around vertical pile. Two different ANNs including multilayer perceptron (with four different learning rules) and radial basis functions neural networks are used for this purpose. The results show that a three-layer normal feed-forward multilayer perceptron with quick propagation (QP) learning rule can predict the scour hole properties successfully.