Eliminating the d=5 proton decay operators from SUSY GUTs

A general analysis is made of the question whether the d=5 proton decay operators coming from exchange of colored Higgsinos can be completely eliminated in a natural way in supersymmetric grand unified models. It is shown that they can indeed be in SO(10) while at the same time naturally solving the doublet-triplet splitting problem, having only two light Higgs doublets, and using no more than a single adjoint Higgs field. Accomplishing all of this requires that the vacuum expectation value of the adjoint Higgs field be proportional to the generator I_{3R} rather than to B-L, as is usually assumed. It is shown that such models can give realistic quark and lepton masses. We also point out a new mechanism for solving the \mu problem in the context of SO(10) SUSY GUTs.