A clinical radiobiological study of the fractionation sensitivity of impaired shoulder movement after postmastectomy radiotherapy is presented. From 1978 to 1980, 163 breast cancer patients received postmastectomy irradiation delivered in 12 fractions, 2 fractions per week, over a period of 37 to 46 days. The total dose was specified either as a maximum absorbed dose of 51.36 Gy, or as a minimum target dose of 36.6 Gy specified at the level of midaxilla. From 1981 to 1982, 66 patients received a minimum target dose of 40.92 Gy in 22 fractions with 5 fractions per week over a period of 29 to 35 days. The treatment technique remained essentially unchanged during the whole period of 1978 to 1982. Maximal flexion and abduction of the arm were measured with unfixed scapula. A relative score, indicating the grade of impaired shoulder movement on a nominal scale from 0–3, was established by comparison with the mobility of the contralateral arm. In addition, working ability and pain at movement and at rest were evaluated. These performance parameters significantly correlated with the grade of impaired shoulder movement. The occurrence of moderately or severely impaired shoulder movement was analyzed using a mixture model incorporating dose-fractionation, latency, and predisposing clinical factors. The risk of shoulder problems was significantly higher among patients above 60 years of age. The analysis pointed to subcutaneous fibrosis as a disposing factor and indicates that postoperative physical exercise programs are beneficial. Arm edema had no statistically significant influence on shoulder performance. The a/,Q ratio was estimated at 3.5 Gy with 95% confidence limits (c.l.) 10.7, 6.21 Gy. The length of time to expression of 90% of the ultimate frequency of moderate and severe shoulder impairment was estimated at 3.9 years with 95% confidence limits 13.1, 4.61 years. The nominal standard dose (NSD) formalism failed to produce isoeffective total doses with respect to impaired shoulder movementin the two fractionation schedules. Although the deleterious effects of larger-than-conventional dose fractions have been documented in a number of clinical series, this study represents the first quantitation of the fractionation sensitivity of a functional clinical endpoint.