The purpose of this overview is to describe radiotherapy retreatment of primary central nervous system tumours from a practical clinical management perspective, including patient selection, choice of radiation technique, dose and fractionation. Useful relief of clinical symptoms and occasionally prolonged survival can follow retreatment. Further analysis of a previously published data set shows that the duration of remission after initial radiotherapy does not correlate with the duration of the remission after retreatment. Also there is no clear relationship between delivered tissue and tumour biological effective dose (BED) and duration of second remission. 'Recovery' of radiation tolerance with time is important and the radiobiological experiments that show this phenomenon have important limitations. To improve the decision as to how much recovery safely occurs with increasing time after radiotherapy, a new mathematical formulation is proposed. This is essentially conservative in its intent, compatible with experimental data sets, and provides a method for tentative calculation of retreatment dose and fractionation. Worked examples are provided of such calculations. As an increasing number of relatively young patients are now retreated, it is important to extend the experimental and human evidence base. A nationally co-ordinated analysis of patients already retreated would be valuable, in order to make future retreatment as safe and effective as possible, with validation of the permissible retreatment schedules for the particular radiation technique used. A national register and task force is proposed to facilitate this.