The radiation reaction from the surface dose to a patient's skin is clinically important as an indicator of tolerance to treatment and is frequently the most visible side-effect from external beam radiotherapy. To quantify surface dose, its build-up was investigated in a solid water phantom over a range of field sizes, source to skin distances, wedge angles and beam shaping conditions. These data were analysed and used to produce an empirical model of the effect of these factors which predicted surface dose to within 5% under test conditions. The effects of changing treatment techniques was investigated by comparing fixed to dynamic wedges, and multileaf collimator (MLC) with alloy blocks. It was found that the MLC reduced the surface dose relative to alloy blocks but that dynamic wedges made no change relative to open fields. Fixed wedges reduced the relative surface dose as a consequence of beam hardening. It is hoped that routine clinical use of the model to calculate skin dose will increase awareness of situations in which the skin sparing effect of modern linear accelerators may be compromised.