Progress in quantitative clinical radiobiology has improved the possibilities for rational design of new radiotherapy schedules. This paper reviews some general problems in calculating the required number of patients in a trial with a radiobiological rationale. Three such rationales are considered: dose escalation, hyperfractionation, and accelerated fractionation. One crucial factor in calculating the size of a trial is the steepness of the dose-response curve for both tumors and normal tissues, and literature data on this are reviewed. It is concluded that fairly large trials, typically comprising 300 or more patients, are necessary, unless efficient stratification of the patients is possible according to the risk for some specific type of recurrence. Such stratification may be possible either according to clinico-pathological characteristics or to the results from predictive assays.