Partial breast irradiation (PBI) is currently under investigation in several phase III trials and, following a recent consensus statement, its use off-study may increase despite ongoing uncertainty regarding optimal target volume definition. We review the clinical, pathological and technical evidence for target volume definition in external beam partial breast irradiation (EB-PBI). The optimal method of tumour bed (TB) delineation requires X-ray CT imaging of implanted excision cavity wall markers. The definition of clinical target volume (CTV) as TB plus concentric 15 mm margins is based on the anatomical distribution of multifocal and multicentric disease around the primary tumour in mastectomy specimens, and the clinical locations of local tumour relapse (LR) after breast conservation surgery. If the majority of LR originate from foci of residual invasive and/or intraduct disease in the vicinity of the TB after complete microscopic resection, CTV margin logically takes account of the position of primary tumour within the surgical resection specimen. The uncertain significance of independent primary tumours as sources of preventable LR, and of wound healing responses in stimulating LR, increases the difficulties in defining optimal CTV. These uncertainties may resolve after long-term follow-up of current PBI trials. By contrast, a commonly used 10mm clinical to planning target volume (PTV) margin has a stronger evidence base, although departmental set-up errors need to be confirmed locally. A CTV-PTV margin >10mm may be required in women with larger breasts and/or large seromas, whilst the role of image-guided radiotherapy with or without TB markers in reducing CTV-PTV margins needs to be explored.