This clinical study was designed to evaluate the safety and efficacy of the sustained release form of dibutryl adenosine-3',5'-cyclic monophosphate (dB-cAMP, bucladesine) placed in the tumor resection cavity at the time of recurrence of the de novo glioblastoma multiforme (GBM) patients. In a randomized prospective manner, 40 patients who were diagnosed as GBM in their first operations were included in this study. Four different therapy protocols were used: First group of 10 patients had tumor resection only. Second group assessed had only systemic chemotherapy as six i.v. infusions of fotémustine after tumor resection. Third group had implantation of bucladesine-loaded biodegradable polymeric sustained release (bcl-SR) pellets while the last group received six i.v. infusions of systemic fotémustine as in the second group in addition to local implantation of bcl-SR pellets. A biodegradable polymer, poly-DL-lactide-co-glycolide with molecular weight of 80000, was used as carrier matrix for the drug with an approximately 4-5 months of release time. Maximal doses of 20 mg of bucladesine with a mean dose of 15.5 mg were implanted. No bone marrow suppression occurred and there were no wound infections as far as the local bucladesine-loaded polymer therapy is concerned. In this randomized prospective trial of local interstitial chemotherapy with long acting bcl-SR did show a statistically significant delay of recurrence on the treatment of GBM patients. Best treatment results obtained from the local bcl-SR + systemic fotémustine treated group in which survival rate estimated by the Kaplan-Meier method was 70% in de novo GBM at 12 months.