This paper focuses on the application of neural networks as an alternative computational tool for the estimation of normalized response spectra for horizontal ground motions with magnitudes MJMA≥5 and hypocentral distance of less than 50 km. The feasibility of using perceptron neural networks in estimating site-specific response spectra and the effects of geophysical properties of site is examined. Two neural network models are proposed for generating normalized response spectra considering the effect of local site conditions. Model 1 was developed with six inputs (magnitude, hypocentral distance, primary wave velocity, shear wave velocity, N-values obtained by Standard Penetration Test (SPT), and density of soil) whereas; model 2 was developed with three inputs (magnitude, hypocentral distance and shear wave velocity). As expected, better performance was obtained from neural network model1 in terms of accuracy and efficiency. Results obtained from this study are very encouraging and has a potential to replace the commonly used regression approach.