Background: Plasma levels of VEGF-A short isoforms (VEGF-A110 and -A121) measured by immunological multiparametric chip technique (IMPACT) were reported to be associated with clinical benefits from bevacizumab (BV) in advanced gastric and pancreatic cancer but not in metastatic colorectal cancer (mCRC). Negative results in mCRC studies might be caused by different sample handling: citrate instead of EDTA and repetition of freeze/thaw. Methods: Blood samples were collected in EDTA before the first-line treatment with BV+mFOLFOX6 or +XELOX for mCRC. Plasma samples were analyzed at Roche Diagnostics Ltd. (Penzberg, Germany) using IMPACT-2 (Roche proprietary multiplex enzyme-linked immunosorbent assay platform). A median value of pVEGF-A was used as a cut-off point to categorize patients (pts) into the low and high pVEGF-A groups. Progression free survival (PFS) and overall survival (OS) between the low and high pVEGF-A groups were compared, using Cox proportional hazards model. We hypothesized that BV-containing treatment extend shorter PFS of pts with high pVEGF-A to that with low pVEGF-A, and estimated a threshold hazard ratio (HR) between them as below 1.15. Results: Among 102 pts enrolled between January 2014 and April 2015, 100 (53 BV+mFOLFOX6 and 47 BV+XELOX) were eligible. Median PFS was 11.4 months [95% CI, 9.5-13.0] and response rate was 64.6 % [range, 53.3-74.9]. pVEGF-A was measured in 97 pts and the median value was 36.8 pg/ml [range, 6.5- 262.2]. The hazard ratios of PFS and OS between the high and low pVEGF-A groups were 1.23 [95%CI, 0.76-1.97, p = 0.40] and 2.47 [95%CI, 1.14-5.36, p = 0.02], respectively. Conclusions: mCRC pts with high pVEGF-A showed shorter PFS than those with low pVEGF-A beyond the predefined threshold (HR 1.15) in BV-containing chemotherapy, suggesting that pVEGF-A could not be a predictive marker for BV efficacy. Clinical trial information: UMIN000012442.