Radioactive substances in the human body can be identified and quantified by gamma spectroscopy using whole body counters. Counting efficiencies needed for calculation of incorporated activities are generally determined from measurements of phantoms simulating shape and density of a human and filled with known activity concentrations. The Cologne whole body counter setup was simulated using the EGSnrc Monte Carlo code system. The simulations did reproduce the spectra and efficiencies from phantom measurements (within +/- 2% for K-40). Variations of the phantom position alongside the stretcher resulted in parabola-shaped courses with efficiency changes of up to 5%. Nuclides which are inaccessible to phantom measurements can be quantified by weighted summation of efficiencies generatedfrom simulation offictitious monoenergetic gamma emitters. For I-131, a strong dependence upon the activity distribution inside the body was observed in simulations with a simplified model of the human body Inclusion of the skeleton in the model had a rather small effect. The efficiency decreases linearly with body length by up to 6% when body mass is kept constant. This has to be taken into account when the activity needs to be determined with high precision. For in vivo counting in the context of radiation protection, however, efficiencies can be deduced with sufficient accuracy from measurements or simulations of simple phantoms.