Background and Purpose : Indications of the existence of long repair half-times on the order of 2–4 h for late-responding human normal tissues have been obtained from continuous hyperfractionated accelerated radiotherapy (CHART). Recently, these data were used to explain, on the basis of the biologically effective dose (BED), the potential superiority of fractionated high-dose rate (HDR) with large fraction sizes of 5–7 Gy over continuous low-dose rate (LDR) irradiation at 0.5 Gy/h in cervical carcinoma. We investigated the optimal fraction size in HDR brachytherapy and its dependency on treatment choices (overall treatment time, number of HDR fractions, and time interval between fractions) and treatment conditions (reference low-dose rate, tissue repair characteristics). Methods and Materials : Radiobiologic model calculations were performed using the linear-quadratic model for incomplete mono-exponential repair. An irradiation dose of 20 Gy was assumed to be applied either with HDR in 2–12 fractions or continuously with LDR for a range of dose rates. HDR and LDR treatment regimens were compared on the basis of the BED and BED ratio of normal tissue and tumor, assuming repair half-times between 1 h and 4 h. Results : With the assumption that the repair half-time of normal tissue was three times longer than that of the tumor, hypofractionation in HDR relative to LDR could result in relative normal tissue sparing if the optimum fraction size is selected. By dose reduction while keeping the tumor BED constant, absolute normal tissue sparing might therefore be achieved. This optimum HDR fraction size was found to be largely dependent on the LDR dose rate. On the basis of the BED NT/TUM ratio of HDR over LDR, 3 × 6.7 Gy would be the optimal HDR fractionation scheme for replacement of an LDR scheme of 20 Gy in 10–30 h (dose rate 2–0.67 Gy/h), while at a lower dose rate of 0.5 Gy/h, four fractions of 5 Gy would be preferential, still assuming large differences between tumor and normal tissue repair half-times and equal overall treatment time. For the same fraction size, an even larger normal tissue sparing can be obtained by prolongation of the HDR overall treatment time. Conclusion : Radiobiologic model calculations presented here aim to demonstrate that hypofractionation in HDR might have its opportunities for widening the therapeutic window, but definitely has its limits. For each specific combination of the parameters, a theoretical optimal HDR fraction size with regard to relative or absolute normal tissue sparing can be estimated, but because of uncertainty in the biologic parameters, these hypofractionation schemes cannot be generalized for all HDR brachytherapy indications.