Background and purpose Small field dosimetry continues to be problematic for the planning and delivery of both circular stereotactic radiotherapy (SRT) fields and intensity modulated radiotherapy (IMRT) fields. Although the separation of head and phantom scatter is primarily of interest in IMRT fields, measurements can often been difficult. However, the set-up in fields formed by stereotactic collimators is more precise and total scatter and head scatter factors can be more easily measured. Phantom scatter factors calculated from these measurements can be extrapolated from the stereotactic situation to the IMRT one. However, the problem of measuring small field head scatter factors must be overcome first. In this work, different techniques were examined to measure small (4–1 cm width) field head scatter factors and determine an optimum phantom and the simplest methodology to ensure set-up accuracy. Methods and materials All measurements were carried out on the 6MV beam of a Varian 600CD. Different phantom diameters, depths and materials were compared and the effects of electron contamination in the lateral and forward directions assessed. Measurements were carried out in fields formed by stereotactic collimators (4–1.25 cm diameter) and in open fields defined by the movable linac collimators (4 × 4 to 1 × 1 cm 2). Results A small plastic top, equal in thickness to the build-up depth and equal in diameter to that of an appropriate solid state detector, was found to be the most suitable phantom. This allows measurements in fields ⩾1 cm, with the detectors investigated. Small differences were found between results in open (standard collimator) fields and in those defined by the stereotactic collimators, but these were likely to be due to differences in beam fluence and in phantom scatter. Conclusion The use of an appropriate small plastic top to measure head scatter factors shows that it is not necessary to preserve lateral electronic equilibrium, nor is electron contamination a problem. Monte Carlo modelling would be useful to investigate the differences further.