We investigated the spatial dependence of the sensitivity of nitrogen measurement with a prompt gamma neutron activation analysis
(PGNAA) system for small animals and developed an analysis procedure that permits the reduction of systematic errors due to
that dependence. The analysis procedure is based on neutron and photon transport calculations performed using the MCNP code
in order to evaluate the sensitivity of the PGNAA facility. The system can be calibrated experimentally using a small number
of phantoms of known size and composition. The calculation approach can then be used to predict responses for animal body
sizes and shapes relatively to those experimentally determined and to include the effect of tissue inhomogeneities. Our calculations
were verified by experimental measurements performed for a set of cylindrical inhomogeneous phantoms. The calculated to experimental
ratios observed were within 6%.