The development of molecular biology over the past thirty years has lead to an explosive growth of knowledge of protein structure and of methods for the design and manufacture of proteins of almost any desired sequence. While this seemingly opens limitless possibilities for the engineering of proteins with novel structures and functions, progress is in reality limited by our inability to predict function from structure, or even structure from sequence. Some progress in the understanding of protein folding and of the determinants of biological specificity has been made, but in essence these remain among the major unsolved problems of molecular biology. Therefore, no contemporary discussion of the subject can give the newcomer to the field a recipe for making a protein that will do exactly what he wants it to do. The best one can hope for is to survey the limits of the known in order to chart a few new inroads into the unknown.