Pattern recognition in proteins has become of central importance in Molecular Biology. Proteins are macromolecules composed of an ordered sequence of amino acids, referred to also as residues. The sequence of residues in a protein is called its primary structure. The 3-D conformation of a protein is referred to as its tertiary structure. During the last decades thousands of protein sequences have been decoded. More recently the 3-D conformation of several hundreds of proteins have been resolved using X-ray crystallographic techniques. Todate, most work on 3-D structural protein comparison has been limited to the linear matching of the 3-D conformations of contiguous segments (allowing insertions and deletions) of the amino acid chains. Several techniques originally developed for string matching have been modified to perform 3-D structural comparison based on the sequential order of the structures. We present an application of pattern recognition techniques (in particular matching algorithms) to structural comparison of proteins. The problem we are faced with is to devise efficient techniques for routine scanning of structural databases, searching for recurrences of inexact structural motifs not necessarily composed of contiguous segments of the amino acid chain. The method uses the Geometric Hashing technique which was originally developed for model-based object recognition problems in Computer Vision. Given the three dimensional coordinate data of the structures to be compared, our method automatically identifies every region of structural similarity between the structures without prior knowledge of an initial alignment. Typical structure comparison problems are examined and the results of the new method are compared with the published results from previous methods. Examples of the application of the method to identify and search for non-linear 3-D motifs are included.