In this paper a comparison of measured and calculated absorbed dose distributions due to 6 and 8 MeV electron irradiation of a thorax phantom consisting of layers of tissue equivalent materials and infinite cylindrical rib inhomogeneities are presented. Dose measurements, carried out in lung equivalent tissue using thermoluminescent dosimeters for the phantom with and without the rib inhomogeneities, are compared with calculations using a typical two-dimensional (2-D) treatment planning system and EGS4 Monte Carlo calculations. The results show that the used 2-D planning system generally reproduces the dose distribution for the phantom without the cylindrical inhomogeneities. As results from the calculations of ratios of the doses in the phantom with ribs to the doses in the phantom without ribs, the dose perturbations are well reproduced by the EGS4 Monte Carlo calculations, but seriously underestimated by the planning system. The deviations are in the same direction for both energies, but are more pronounced for 6 MeV electrons. The disagreement between the results of the planning system on one hand and the Monte Carlo calculations and experiments on the other hand for the multirib system used is examined further by calculating dose distributions for a phantom containing only one infinite rib. The calculations show that the 2-D treatment planning system generally yields broader and less profound minima in the dose distribution directly beneath the inhomogeneities, with deviations more prominent at lower energies. In addition to the limitation of the central ray approximation, it is suggested that these discrepancies may be due to a too large electron pencil beam spread resulting from the implementation of a range straggling modification function of limited validity at the depths beyond dose maximum for these energies and these tissues.