Developing optimal radiation-free central nervous system prophylaxis is a desirable goal in acute lymphoblastic leukemia, to avoid the long-term toxicity associated with cranial irradiation. In a randomized, phase II trial enrolling 145 adult patients, we compared intrathecal liposomal cytarabine (50 mg: 6/8 injections in B-/T-cell subsets, respectively) with intrathecal triple therapy (methotrexate/cytarabine/prednisone: 12 injections). Systemic therapy included methotrexate plus cytarabine or L-asparaginase courses, with methotrexate augmented to 2.5 and 5 g/m(2) in Philadelphia-negative B- and T-cell disease, respectively. The primary study objective was the comparative assessment of the risk/benefit ratio, combining the analysis of feasibility, toxicity and efficacy. In the liposomal cytarabine arm 17/71 patients (24%) developed grade 3-4 neurotoxicity compared to 2/74 (3%) in the triple therapy arm (P=0.0002), the median number of episodes of neurotoxicity of any grade was one per patient compared to zero, respectively (P=0.0001), and even though no permanent disabilities or deaths were registered, four patients (6%) discontinued intrathecal prophylaxis on account of these toxic side effects (P=0.06). Neurotoxicity worsened with liposomal cytarabine every 14 days (T-cell disease), and was improved by the adjunct of intrathecal dexamethasone. Two patients in the liposomal cytarabine arm suffered from a meningeal relapse (none with T-cell disease, only one after high-dose chemotherapy) compared to four in the triple therapy arm (1 with T-cell disease). While intrathecal liposomal cytarabine could contribute to improved, radiation-free central nervous system prophylaxis, the toxicity reported in this trial does not support its use at 50 mg and prompts the investigation of a lower dosage. (clinicaltrials.gov identifier: NCT-00795756).