Experimental verification of the difference between absorbed dose in tissue and the collision fraction of kerma requires precise knowledge of the absorbed dose curve, particularly in the build-up and build-down regions. A simple method for direct measurement of contamination-free build-up for 60Co, which should also be applicable for most of the photon energies commonly employed for treatment, is presented. It is shown that the contribution from air-scattered electrons to the surface dose may be removed by extrapolating measurements of build-up to zero field size. The remaining contribution to contamination from the collimators and other source-related hardware may be minimised by measuring these build-up curves sufficiently far from the source. These results were tested by measuring the build-up using a magnet to sweep scattered electrons from the primary photon beam and by measuring the surface dose in the limit of an evacuated beam path. The relative dose at zero depth in polystyrene was found to be approximately 8.9+or-0.3% of the dose at the depth of maximum build-up.