In our previous study, the simulation of a cyclotron-based neutron field for boron neutron capture therapy (BNCT) using a (p,n) spallation source with the MCNPX code was validated through measurements of the neutron energy spectrum behind the moderator assembly and the thermal neutron distribution in an acrylic phantom using reaction rates of 198 Au. These validations showed that the simulation generally well reproduced the measurements. However, some discrepancies between the measurements and the calculation remained for clinical trials. In this paper, we investigated the influences of neutron source spectrum and thermal neutron scattering law data in the simulation to resolve those discrepancies. We also compared measured and calculated neutron doses behind the moderator assembly with results obtained using a tissue equivalent proportional counter. We clarified that the neutron source spectrum calculated using the LA150 data led to the overestimation of high-energy neutrons in a phantom, but this overestimation did not significantly affect the neutron dose distribution in a phantom, because a dominant part of the absorbed dose is due to neutrons of energies below 1 MeV. The study of the influence of neutron scattering law data in a phantom also indicated that the use of selected S(,) data led to an improvement in the simulation of thermal neutron behavior.