This paper presents results from absorbed dose calculations carried out usi
ng the Monte Carlo N-particles (MCNP) code in the Portuguese Gamma Irradiat
ion Facility. The computer running time was chosen in such a way that the s
tatistical errors were always less than 5%. Dose measurements using Amber P
erspex and ceric-cerous dosimeters were performed in order to validate the
calculations. Simulated and experimental data show good agreement even at p
ositions with a high gradient dose. This fact indicates that the model is r
eliable. Consequently: (a) the code was used in a correct way; (b) source a
nd geometry of the installation were well defined; and (c) dimension, compo
sition and density of the product were correctly established. The amount of
experimental work with dosimeters can thus be reduced. The results show th
at Monte Carlo simulations can be used as a predictive tool of dose measure
ments in an irradiation plant. However, some validation measurements must b
e performed. (C) 2000 Elsevier Science Ltd. All rights reserved.