Aim: To describe patterns of radiotherapy retreatment by cancer type in order to develop methods of modelling retreatment for better service planning and benchmarking. Materials and methods: We examined an institutional database of all patients who received their first megavoltage radiotherapy for any type of cancer at the Liverpool and Macarthur Cancer Therapy Centres over the period 1997-2006. The database contains patient demographic data, dates, treatment sites and doses of radiotherapy. The treatment sites were entered as free text and were recoded into a limited number of sites; primary site, bone, brain, soft tissue, other and multiple, so that retreatment sites could be grouped for stratification and analysis. Multiple treatment sites that were part of a single treatment to a primary (for example, primary site and nodes) were recoded as 'primary'. The total retreatment ratio was defined as the number of episodes of radiotherapy divided by the number of cases in the cohort. Results: In the period 1997-2006, 7853 patients had received 9859 episodes of radiotherapy, giving a total retreatment ratio of 26 per 100 patients. In total, 6524 patients (83%) received only one episode of treatment and 1329 received two or more episodes of treatment. The average number of retreatments was 0.26. The tumour type with the highest mean number of retreatments was myeloma (1.05), followed by unknown primary (0.41), lung (0.34) and melanoma (0.32). The lowest mean number of retreatments was for the brain (0.06). The median time between treatment episodes was longest for breast cancer (12.5 months), then colorectal cancer (12.3 months), head and neck cancers (9.0 months), lung cancer (3.8 months) and prostate cancer (5.9 months). Conclusion: The retreatment ratio for the cohort was affected by the length of follow-up and by the tumour type. The mean number of retreatments varied from 0.06 for central nervous system malignancy to 1.05 for myeloma, with an average of 0.026.