Background: There is no standard chemotherapy defined in PCNSL. Elderly patients (pts) are not candidates for whole brain radiotherapy and therefore establishing an optimal MTX-based regimen is crucial. This prospective multicenter study conducted in 13 French institutions tested two promising MTX-based chemotherapy regimens in elderly pts with newly diagnosed PCNSL. Methods: Pts with histologically confirmed newly diagnosed PCNSL with age ≥60 and KPS ≥40 were stratified by institution and KPS, then randomized to receive three 28-day cycles of MTX (3.5 g/m² D1 and D15) and TMZ (100-150mg/m2 D1-5 and 15-19) [MT arm] or 3 cycles of MTX (3.5 g/m² D1 and D15), procarbazine 100mg/m² (D1-7), vincristine (1.4mg/m2 D1 and 15), followed by cytarabine consolidation (3g/m²/d X2d) [MPV-A arm]. Neither arm included radiotherapy; prophylactic G-CSF and standardized corticosteroids (methylprednisolone 60mg/d D1-5) were given to both arms. The primary endpoint was PFS (one-stage Fleming design; α= 5%; β=10%). Evaluations included neuropsychological testing and quality of life. Results: Accrual has been completed (7/2007- 3/2010), with 98 pts randomized and 95 analyzed (MT: 48 pts; MPV-A: 47). Pre-treatment characteristics were well balanced between the two arms (all pts: median age=72- range 60-85; median KPS= 70; range 40-100). In the MPV-A arm, the CR rate = 62% (vs 45% in MT arm [p=0.11]), objective response rate= 82% (vs 71%; p=0.23), median PFS= 9.5m (vs 6.1m; HR= 1.14- 95% CI [0.72 ; 1.81]; p=0.6) and median OS= 31m (vs 13.8m; HR= 1.4 - 95% CI [0.84 ; 2.34]; p=0.2). The incidence of grades 3-4 toxicities was 72% in the MPV-A vs 71% in the MT arm. Abnormal liver function test was the most common toxicity (MPV-A: 18 pts; MT: 21). Baseline cognitive impairment (MMSE >24 vs ≤24) predicted OS (p=0.04). Conclusions: This is the first randomized PCNSL study testing two different MTX-based combination regimens. In this elderly population, toxicities were frequent but similar in both arms, and all efficacy endpoints tended to favor the MPV-A arm. The MPV-A regimen is recommended for further development in PCNSL. Clinical trial information: NCT00503594.