During radiation therapy treatments, neutron contamination can be a source of unwanted radiation dose to the patient and medical personnel. Accurate cross-section data is needed to characterize the neutron contamination in medical accelerators using Monte Carlo methods. In this study, a comparison of the photoneutron yields using the default LA150U and the Chinese Nuclear Data Center (CNDC) photonuclear cross sections was performed. Thick tungsten plates, each of 0.125-cm thickness (one-third radiation length), were directly irradiated by an electron beam in MCNPX. In order to match typical radiation therapy energy ranges, the energy distribution of the electron beam was modeled as a Gaussian distribution with a mean energy of 18.3 MeV and a 3% full-width at half-maximum. The photoneutron yield using the LA150U is consistently [approximately]12 to 17% higher than those from the CNDC data for each target thickness. The average photoneutron energy difference between the two cross-section libraries ranged from 3 to 42%. No major differences were seen between relative neutron fluences per solid angle for the two cross-section libraries. The discrepancies between the datasets provided above can be attributed to the oversimplification of using the default LA150U 184W cross section for all other naturally occurring isotopes of tungsten. Therefore, the lack of cross-section data in the LA150U library is a definite concern when using MCNPX to determine secondary neutron production in a medical accelerator room since a majority of contamination neutrons are produced in tungsten components.