Fusion of very neutron rich nuclei may be important to determine the composition and heating of the crust of accreting neutron stars. Fusion cross sections are calculated using time-dependent Hartree-Fock theory coupled with density-constrained Hartree-Fock calculations to deduce an effective potential. Systems studied include 16O+16O, 16O+24O, 24O+24O, 12C+16O, and 12C+24O. We find remarkable agreement with experimental cross sections for the fusion of stable nuclei. Our simulations use the SLy4 Skyrme force that has been previously fit to the properties of stable nuclei, and no parameters have been fit to fusion data. We compare our results to the simple S\~{a}o Paulo static barrier penetration model. For the asymmetric systems 12C+24O or 16O+24O we predict an order of magnitude larger cross section than those predicted by the S\~{a}o Paulo model. This is likely due to the transfer of neutrons from the very neutron rich nucleus to the stable nucleus and dynamical rearrangements of the nuclear densities during the collision process. These effects are not included in potential models. This enhancement of fusion cross sections, for very neutron rich nuclei, can be tested in the laboratory with radioactive beams.