To evaluate dose errors in balloon catheter brachytherapy of breast cancer due to inhomogeneities, such as iodine-containing radiographic contrast medium in the balloon, the lack of scattering medium, and the low density of lung that are not considered by commercial treatment planning systems (TPS). By accounting for these inhomogeneities in breast/lung phantoms, Monte Carlo simulations were performed to calculate doses in the breast and lung. Doses were also calculated by a commercial TPS. The Monte Carlo doses and the TPS doses were compared along the transverse and longitudinal axes of the source. The Monte Carlo doses were lower by 4-10% on the prescription line than the TPS doses, depending on the concentration (5-25% by volume) of the contrast medium, and on the direction from the source. The lack of scattering medium around the breast contributes to the differences more than the attenuation by the contrast medium. Attenuation contributed approximately 1.0-4.8% at the concentrations investigated in this study. Current treatment planning systems, which assume a source in a large homogeneous water-equivalent medium, significantly overestimate doses in breast brachytherapy. © 2004 Elsevier Inc.