This paper examines how physical factors such as the depth of the excision cavity below the skin surface and its distance from the underlying lung may impact upon the choice between the use of an electron field (from a Varian Clinac 1800) or an implant (using a double plane of iridium-192 wires). Data have been derived from a phantom dosimetry experiment simulating different permutations of breast size and depth of excision cavity. An anthropomorphic female phantom with two different-sized wax breast phantoms has been used to simulate the clinical circumstances envisaged and isodose distributions have been estimated from an array of TLD readings. The biological significance of doses measured in skin and lung have been examined using the linear-quadratic (LQ) model. Both on physical and biological grounds, the results favour the implant under the experimental conditions adopted. The use of the electron beam to definitively treat the excision cavity (omitting breast tangents) to 60 Gy would result in unacceptable late effects in the skin and an observable incidence of pneumonitis if the excision cavity were near the chest wall. Small carcinogenic risks, particularly to the lung, are apparent with each modality, but may be moderated for the iridium-192 implant by a reduced carcinogenic potential associated with low-dose rate radiation. The use of the newer radionuclides iodine-125 and samarium-145, with less penetrating gamma ray emissions, might be preferred to iridium-192 from the point of view of bronchial carcinogenesis if definitive treatment of the excision cavity became widespread practice.