AbstractNeural networks are powerful computational tools and have been applied in various ways. In this work a neural network is used to solve a pattern classification problem encountered in biochemistry. One of the major topics of research in molecular biology is the prediction of functional properties of biomacroinoleculcs from their sequence data. A radial basis function (RBF) network is presented for predicting the trausmembrane regions of membrane proteins. A comparison between the results of prediction by the RBF network and a popular conventional method along with real data show the superiority of this approach to prediction of transmembrane regions.