Bevacizumab may improve outcomes of patients with breast cancer, but the absence of an established biomarker hampers patient selection and researchers´ ability to demonstrate a clear survival benefit. Its putative target, circulating VEGF-A, emerged as the main candidate and we sought to identify the relationship between VEGF-A levels and outcomes through systematic review. We searched electronic databases and meeting proceedings for randomized controlled trials (RCTs) comparing the addition of bevacizumab to standard chemotherapy for breast cancer. RCTs were included if outcomes were presented separately according to VEGF-A plasma levels. Random-effects model were applied to calculate the pooled hazard ratios for progression-free survival, event-free survival (EFS), comprising disease recurrence, progression or any-cause death, and overall survival (OS), with respective confidence intervals (95 % CI). High and low VEGF-A levels subgroups followed each trial definition, and results were compared using the interaction test. Heterogeneity was calculated using χ (2) test (I (2)). Three trials enrolled a total of 3748 patients. 1713 patients had baseline VEGF-A levels in plasma available for assessment and were included. One trial added bevacizumab in the adjuvant setting (N = 2591) and two on first-line metastatic disease with taxane-based therapy (N = 1160) There was no interaction between VEGF-A levels and study setting (adjuvant vs. first line therapy). Bevacizumab improved PFS of patients with above median VEGF-A plasma levels (HR 0.56; 95 % CI 0.43-0.73; P < 0.001; I (2) = 0 %), but not of those with below median VEGF-A levels (HR 0.89; 95 % CI 0.68-1.15; P = 0.37; I (2) = 0 %), with relevant differences between these two groups, P-for interaction = 0.02. The same happened with EFS (VEGF-A above median HR 0.62; 95 % CI 0.39-0.79; P < 0.001; I (2) = 11 %; below median HR 0.89; 95 % CI 0.71-1.14; P = 0.98; I (2) = 17 %; P-for interaction = 0.03). OS data were not available. VEGF-A level is a reasonable candidate biomarker for bevacizumab in the treatment of breast cancer. Further studies have to confirm its surrogacy in overall survival and in other scenarios including other anti-angiogenic therapies.