A multicentre phase II trial of stereotactic body radiotherapy for T1N0M0 non-small cell lung cancer was initiated in Japan as the Japan Clinical Oncology Group trial (JCOG0403). Before starting the trial, a decision was made to evaluate the treatment machine and treatment planning in participating institutions to minimize the variations of the prescription dose between the institutions. We visited the 16 participating institutions and examined the absolute dose at the centre of a simulated spherical tumour of 3.0 cm diameter in the lung using the radiation treatment planning systems in each institution. A lung phantom for stereotactic body radiotherapy (SBRT) was developed and used for the treatment planning and film dosimetry. In the JCOG radiotherapy study group, the no model-based calculation algorithm or the model-based calculation algorithm with a dose kernel unscaled for heterogeneities were selected for use in the initial SBRT trials started in 2004, and the model-based calculation algorithm with a dose kernel scaled for heterogeneities was selected for the coming trial. The findings of this study suggest that the clinical results of lung SBRT trials should be carefully evaluated in comparison with the actual dose given to patients.