The immunological therapy of cancer has been proposed in a number of neoplasms (Borden, Sondel, 1989; Foon, 1989; Rosenberg, 1992) and has recently been adopted in the treatment of Central Nervous System (CNS) tumors in combination with conventional surgical and radiotherapeutical approach. In this context, loco-regional administration of immunomodulating agents (for instance in post-surgical cavity) allows to achieve much higher in situ concentrations than by systemic route. Since these treatments have potential adverse effects, careful assessment of clinical and immunological parameters in phase I trials is needed. CNS tumors disseminating via Cerebrospinal Fluid (CSF) pathways offer a stimulating opportunity for intrathecal immunotherapy. In this context, alpha-IFN and IL2 (alone or in combination with LAK cells) have been employed either loco-regionally or intrathecally (Merchant, Mc Vicar, Merchant & Young, 1992; Schiller, Hank, Storer, Borchert, Moore, Albertini, Bechhofer, Wesley, Brown, Bastin & Sondel, 1993). The rationale for the use of both these substances includes the known anti-tumor action of alpha-IFN (Mahaley, Urso, Whaley, Blue, Williams, Guaspari & Selker, 1985; Nagai, 1988) and the ability of r-IL2 to generate activated cells effective in lysing tumor cell targets (Hayes, Moore, Pierz, Chen, Da Rosso, Nirenberg & Allen, 1993). We treated 3 patients (2 affected by disseminating cerebellar medulloblastoma, 1 by disseminating thalamic glioblastoma) by intrathecal r-IL2 via recervoir. In the first 2 patients, this treatment was preceded by alpha-IFN (also intrathecally). Monitoring of immunological effects of the treatment schedule involved kinetics of CSF and serum TNF-alpha, IL2s and IL2R during the first day of r-IL2 treatment, as well as on day +2 and +4 of both r-IL2 cycles, and assessment of CSF cells, protein and CSF and PB NK cell activity and CD3-CD56+ cells during the course of all treatment cycles. We also assessed clinical and neuroradiological effects of immunotherapy.