Scheme 1 summarizes some of what we have learned from this study of non-viral protein substrates of the HIV proteases. Many of these findings contradict the current understanding of protease specificity. P1-P1' amino acids need not be bulky or hydrophobic and residues at these positions may be even less important than those in flanking positions (e.g., Glu at P2') in dictating the course of hydrolysis. Thus, the pattern of amino acids over the whole binding region must be considered in predicting what will or will not be a substrate of these enzymes and, although we are beginning to understand selectivity at the level of primary structure, a detailed explanation of their specificity is yet to be forthcoming. Nevertheless, studies of this kind find useful application in the design of inhibitors of HIV proteases that will, hopefully, be of value in treatment of AIDS.