Two kinds of quantum mechanical treatment are needed in chemistry: First, a treatment accurate enough to provide definitive solutions of chemical problems, in particular, the prediction of reactions and reaction mechanisms, without reference to experiment. As yet, treatments of this degree of accuracy are restricted to a few small atoms and molecules. Second, a treatment that chemists can use as a practical aid to their own research, on the same basis as mass spectrometry or NMR. Here, the need has been largely met by the series of progressively better semiempirical methods, in particular, those developed by our group. Since only the best and most expensive ab initio procedures are superior to ours and since the computing time they need is far greater, their use as a practical aid in chemistry is restricted. Our procedures, therefore, serve as a useful supplement to ab initio ones in areas where the latter cannot be applied effectively, as the leading ab initioists now fully recognize. Since the semiempirical approach is capable of almost unlimited further development, this situation is likely to continue in the foreseeable future. 0 1992 John Wiley & Sons, Inc.