Background: We have previously showed that a high MMP2, and to a lesser extend a low MMP9 plasma levels were associated to a high response rate, a prolonged PFS and OS in recurrent GB treated with bevacizumab (BEV), but not with cytotoxic agents, (Tabouret and col. Neuro-Oncol 2013). In order to further explore the optimal timing of bevacizumab administration, we analyzed potential differences of MMP2/MMP9 plasma levels and tumor RNA in patients with newly diagnosed and recurrent GB. Methods: Plasma was collected before radiotherapy in newly diagnosed GB patients (pts) (Pop ND) and in a distinct population of pts at the time of recurrence (Pop RD). MMP2 and MMP9 plasma levels were assessed using ELISA. In a third population with paired initial and recurrent GB tumors (Pop NDT and RDT respectively), MMP2 and MMP9 RNA were analyzed using quantitative RT PCR. Correlations were analyzed using the Mann Whitney U test, the Spearman corelation and the T-test. Results: MMP plasma levels were tested in 44 ND and 76 RD pts. Mean levels of ND and RD MMP2 levels were 196.4 pg/ml (± 9.6) and 218.2 pg/ml (± 9.6). Mean levels of ND and RD MMP9 levels were 365.4 pg/ml (± 50.7) and 302.0 pg/ml (± 34.9). No significant difference was observed between ND and RD plasma levels of MMP9 and MMP2. In ND pts, no correlation was found between MMP2 and MMP9 plasma levels, neither between these plasma markers and KPS or age. In NR pts, MMP2 and MMP9 plasma levels were inversely correlated (p<0.001). Paired MMP9 and MMP2 RNA were available in NDT and RDT for 29 pts. No difference was observed between NDT and RDT expressions of MMP9 and MMP2. At initial diagnosis, MMP9 and MMP2 were 95.7-fold and 15.9-fold over-expressed compared with normal brain. At recurrence, MMP9 and MMP2 were 80.6-fold and 25.9-fold over-expressed. MMP2 and MMP9 RNA were correlated in NDT (p=0.001) but not in RDT. Conclusions: In this retrospective study, the prebiomarkers of BEV activity, MMP2 and MMP9, did not show significant change between GB initial presentation and recurrence both for plasma levels and tumor RNA. These results may suggest that BEV could be similarly active for pts with newly diagnosed and recurrent GB. Adequate prospective studies are warranted.