Background: Therapy targeting VEGF or EGFR demonstrated clinical activity in combination with chemotherapy (CT) in mCRC but monoclonal antibodies cannot be associated. The DREAM trial compares a maintenance therapy (MT) with bev +/- EGFR tyrosine kinase inhibitor erlotinib (E) after a first-line Bev-based induction therapy (IT) in pts with mCRC. Methods: Pts with previously untreated and unresectable mCRC were eligible. After a Bev-based IT with FOLFOX or XELOX or FOLFIRI, pts without disease progression were randomized to MT between Bev alone (Bev 7.5 mg/kg q3w; arm A) or Bev+E (B 7.5 mg/kg q3w, E 150 mg/day continuously; arm B). Pts were treated until progression or unacceptable toxicity. The primary endpoint was PFS on MT. Results: The study enrolled 700 pts from 01/2007 to 11/2011 in 3 countries (France, Canada, Austria). 446 (63.7%) pts were randomized for MT (arm A, N=224; arm B, N=222). Among the 446 randomized pts, IT regimen was FOLFOX-Bev in 265 pts (59.4%), XELOX-Bev in 135 pts (30.3%), and FOLFIRI-Bev in 46 pts (10.3%). Baseline characteristics of randomized pts were (arm A/B): ECOG PS 0, 60% in both arms; normal LDH level 47%/49%; normal alkaline phosphatase level 48%/50%; synchronous metastasis 83%/82%. The median no of MT cycles was 6 in both arms. With a median follow-up of 31.0 months, 327 PFS events were observed. Median MT-PFS were 4.6 m in arm A vs 5.8 m in arm B (HR 0.73 [95%CI: 0.59-0.91], P=.005). Median PFS from inclusion were 9.2 m vs 10.2 m. During MT, in arm A vs arm B, grade 3-4 diarrhea (<1% vs 9%) and grade 3 skin toxicity (0% vs 19%) were the main differences in toxicity. Severe adverse events from randomization related to B or E were 6 in arm A and 7 in arm B. Overall survival is not mature. Conclusions: The addition of erlotinib to bevacizumab after induction therapy significantly improves the duration of maintenance PFS, following induction with first-line chemotherapy plus bevacizumab, in patients with unresectable metastatic colorectal cancer.