A computational model using an MCNPX version 2.6.0 code and a leg voxel phantom was previously constructed and validated against the in vivo measurements of the United States Transuranium and Uranium Registries (USTUR) case 0846 leg. Using the MCNPX model, different simulation scenarios of 241Am distribution in the bones and tissue material of a leg were performed, and their effects on the detection efficiency and activity calculation were examined. The purpose of this work is to ensure and increase the simulation sensitivity of real contaminated human bones and reduce the simulated efficiency error associated with the distribution of 241Am activity within the leg bones when using a high purity germanium [HP(Ge)] detector. The results showed that the simulated detection efficiency obtained from the uniform distribution of 241Am in the leg bones was underestimated by a factor of up to 0.3 compared with the measured and simulated detection efficiency obtained from the non-uniform distribution of 241Am in different sections of the leg bones. The p-value of a one-way analysis of variance (ANOVA) F-test among the mean values of the simulated detection efficiencies was calculated and provided evidence of a significant difference. The uncertainty in the bone activity estimate could be quite large (25% to 30%) if calibration of detection efficiency is based on assuming a uniform distribution of 241Am in the phantom to estimate the USTUR case 0846 leg activity. It is therefore recommended that during calibration of detectors, a non-uniform distribution of 241Am in different sections of the bones should be used rather than a uniform distribution. Additionally, an assumption of a uniform distribution of 241Am will simulate 241Am activity deposited in the leg bones of a real contamination case inadequately.