Patients with head and neck cancer were randomized in a double-blind manner to receive 40 mg/day prednisone (n = 32) or placebo (n = 34). The prescribed daily tumour dose was 1.6 Gy twice daily with an interfraction interval of > or =6 h. Treatment interruptions were at the discretion of the radiation oncologists. The primary study end-point was the total duration of treatment. A mean 4.4 day reduction was observed in the total elapsed treatment duration for the prednisone arm, 34.3 versus 29.9 days (P = 0.013), which paralleled a significantly lower median total tumour and biologically effective dose (BED) Gy10 in this arm. A trend favouring shorter treatment interruptions in the prednisone arm was also found, but not a reduction in the intensity or duration of mucositis.