The present status of the phenomenological neutron optical potential is briefly reviewed, with particular attention to the dependence on the nuclear asymmetry. It is shown that the conventional parametrization is inadequate and some possible improvements are discussed. These improvements are of two types: firstly a more flexible global parametrization and secondly some ways of incorporating data relating to each particular nucleus. Two ways of doing this are proposed, using the RMS matter radius to improve the real potential and the nuclear level density to improve the imaginary potential.