The time-dependent radiation transport for a demonstration scale liquid-metal-cooled fast breeder reactor that has undergone a severe loss of sodium coolant is calculated with both a discrete ordinates and a diffusion theory solution for the real neutron flux shape. It is found that diffusion theory underpredicts reactivity levels by about $6 when compared to discrete ordinates. It is also found that the use of an initial adjoint neutron flux throughout the transient as a reactivity weighting function could seriously underpredict reactivity levels for a severely degraded reactor core. In both cases, there was an immediate termination of the excursion. The uncertainty of being limited to two fuel fields for an end of equilibrium cycle reactor core in SIMMER-II during the transient was greater than that due to microscopic cross-section shielding factor iteration and interpolation schemes. Fifty-energy-group reactivity coefficients were best duplicated in collapsing to a ten-energy-group set with an entire reactor integrated bilinear neutron energy flux spectrum.