A prospective double blind randomized trial comparing two different dose schedules of continuous steroid coverage during prophylactic cranial radiotherapy (CRT) in leukemic children was conducted to find out the optimum dose to be prescribed to reduce the incidence of Somnolence Syndrome (SS). Between April 1994 and February 1996, 32 patients with acute lymphoblastic leukemia received CRT of 18 Gy in 10 fractions. Patients were randomized to receive oral dexamethasone of 2 or 4 mg/m(2) during radiotherapy. The diagnosis of SS was made clinically based on symptoms of somnolence. All patients were followed for a minimum of 8 months. The overall incidence of SS was 40%. The development of SS was steroid dose dependent. In low dose steroid arm the incidence of SS was 64.3% (9/14), compared to 17.6% (3/17) in high dose arm with statistically significant difference (P = 0.008). The median time to development of SS was 4 weeks. The most common symptom of SS was drowsiness followed by anorexia, headache, nausea, vomiting, decreased activity, irritability, fever and ataxia, respectively. The duration of symptoms ranged from 2 to 14 days. The development of SS was not related to the presence of acute reactions, age at the time of CRT and sex. In all cases the symptoms subsided completely and spontaneously. Our results suggest that steroid coverage at a dose of 4 mg/m(2) during CRT reduces the incidence of SS. However, a multicentric prospective randomized trial is needed to determine the role and the optimal dose of steroid. (C) 1998 Elsevier Science Ireland Ltd. All rights reserved.