Planetary thermal neutron fluxes provide a sensitive proxy for mafic and feldspathic terranes and are also necessary for translating measured gamma-ray line strengths to elemental abundances. Both functions require a model for near-surface temperatures and a knowledge of the dependence of thermal neutron flux on temperature. We have explored this dependence for a representative sample of lunar soil compositions and surface temperatures using the Monte Carlo N-Particle Code (MCNP™)(MNCP is a trademark of the Regents of the University of California, Los Alamos National Laboratory). For all soil samples, the neutron density is found to be independent of temperature, in accord with neutron moderation theory. The thermal neutron flux, however, does vary with temperature in a way that depends on Δ, the ratio of macroscopic absorption to energy-loss cross sections of soil compositions. The weakest dependence is for the largest Δ (which corresponds to the Apollo 17 high-Ti basalt in our soil selection), and the largest dependence is for the lowest Δ (which corresponds to ferroan anorthosite, in our selection). For the lunar model simulated, the depth at which the thermal neutron population is most sensitive to temperature is ~30 g cm-2. These simulations were compared with the flux of thermal neutrons measured using the Lunar Prospector neutron spectrometer over the lunar highlands using a subsurface temperature profile that varies with latitude, λ, as Cos1/4λ. Model results assuming equatorial temperatures of 200 and 250 K are in reasonable agreement with measured data. This range of equatorial temperatures is not inconsistent with the average temperature measured below the diurnal thermal wave at the equator, Tmeas = 252 ¿ 3 K <Langseth and Keihm, 1977>.