The shape of the measured photon build-up curve and the surface dose are strongly dependent on the chamber geometry and side wall density. This study examines the response at the phantom surface to a 6 MV beam for the NACP and MARKUS chambers and also for the recently available commercialized ATTIX chamber; and checks the consistency of a published correction term applied to these chambers by comparison of the corrected data sets. The MARKUS and ATTIX chamber results of surface dose versus field size agree to within less than 0.5% over the field size range, whereas the NACP results show an average discrepancy of 22.1% (absolute percentage points), although the slope of the dose versus field size plot is very similar for all three chambers. It is confirmed then that the NACP electron chamber is not optimal for photon dosimetry at an air-tissue interface, and that the chamber of choice from fabrication material and geometrical considerations is the ATTIX solid water chamber.