The effect of head phantom size on the 10B and 1H[n,gamma]2H dose distributions for a broad epithermal neutron radiation field generated by an accelerator-based epithermal neutron source for boron neutron capture therapy (BNCT) have been studied. Also two techniques for calculating the absorbed gamma dose from a measured gamma-ray source distribution are compared: a Monte Carlo technique, which is well accepted in the BNCT community, and a Point Kernel technique. The count-rate distribution in the central plane of three rectangular parallelopiped head water phantoms irradiated with an epithermal neutron field was measured with a boron trifluoride (BF3) detector. This epithermal neutron field was produced at the Ohio State University Van de Graaff Accelerator Facility. The 10B absorbed dose and the gamma-ray source have the same distribution in the head phantom as the BF3 count-rate distribution. The absorbed gamma dose from the measured source distribution was calculated using MCNP, a Monte Carlo code, and QAD-CGGP, a Point Kernel code. The most pronounced effect of phantom size on 10B absorbed dose was on the dose rate at the depth of maximum dose, dmax. An increase in dose rate at dmax was observed with a decrease in phantom size, the dose rate in the smallest phantom being larger by a factor of 1.4 than the dose rate in the largest phantom. Also, dmax for the phantoms shifted deeper with a decrease in phantom dimensions. The shift between the largest and the smallest phantoms was 6 mm. Finally, the smaller phantoms had lower entrance 10B dose as a percent of the dose at dmax, or better skin sparing. Our calculations for the gamma dose show that a Point Kernel technique can be used to calculate the dose distribution as accurately as a Monte Carlo technique, in much shorter computation times.