Seismic evaluation of existing facilities and design of earthquake-resistant structures requires estimation of strong ground motion parameters. Ground motion prediction equations are derived for estimating strong ground motion parameters. Traditionally, ground motion prediction equations are developed using statistical regression analyses. This book presents a very promising application of the powerful and versatile neural network approach to the derivation of ground motion prediction equations using Japanese earthquake records and site characteristics. Multi-layer perceptron neural network models with back-propagation learning scheme have been developed to predict strong ground motion parameters that are of primary significance in earthquake engineering using the actual seismic data without any simplification and assumptions. Statistics of the results presented indicate that artificial neural network is capable of learning and representing the local variations quite accurately. This book is intended to help structural engineers in selection of appropriate characterization of ground motion for engineering applications that are used in seismic analysis and design.