A simulation study of a PGNAA measuring arrangement with a252Cf neutron source for the characterization of cement raw materials was carried out using the MCNP code to investigate the effect on the system response of varying the bulk density and the water contentvw of samples of fixed dry composition. The source is placed at the centre of a lead cylinder of extemal radiusRPb. This cylinder is enclosed in a coaxial cylinder of polyethylene moderator of extermal radiusRin. The sample material is confined to the space between an outer cylindrical surface of radiusRout and the moderator cylinder. The first series of simulation runs was carried out for different sets of values ofRin,Rout and , and constantvw. The system specific responseS (count rate per wt%) shows a broad maximum aroundTM=Rin-RPb=4 cm andTS=Rout-Rin=8.5 cm and a dependence on that is almost linear in the region of the maximum. In a second series of runs the dependence ofS onvw was studied for a geometry corresponding to a real experiment described elsewhere and for a geometry for which the density saturation effect is already apparent. It is shown that when the sensitivity of the system is optimized both andvw must be used as calibration parameters while in the low-sensitivity design regionvw can be used as the only calibration parameter.