Anti-vascular endothelial growth factor therapy (VEGF) is an important new treatment modality in oncology. We sought to determine the efficacy and safety of the humanized monoclonal anti-VEGF antibody, bevacizumab, and vinorelbine as treatment for refractory breast cancer and to explore the role of plasma VEGF as a predictor of treatment outcome. Eligible patients had received one or two prior chemotherapy regimens for metastatic breast cancer or recurred within 12 months of adjuvant therapy and had measurable disease and adequate end-organ function. Patients received bevacizumab 10 mg/kg every 2 weeks, and vinorelbine each week, until tumor progression or prohibitive toxicity. Plasma VEGF was measured at baseline. Among 56 women treated on protocol, bevacizumab and vinorelbine yielded a 34% response rate (95% confidence interval, 22-48%) and median time to progression of 5.5 months. Activity was observed regardless of tumor hormone receptor status or type or extent of prior chemotherapy. Side effects included uncomplicated neutropenia, hypertension, nasal congestion/epistaxis, and neuropathy, consistent with well-described side effects of the respective agents. Three patients had impaired wound healing following surgical procedures. There were only rare instances of thrombosis or clinically significant proteinuria. Lower levels of baseline VEGF were associated with longer time to progression. Bevacizumab and vinorelbine are well tolerated and effective as treatment for refractory breast cancer. Plasma VEGF warrants further evaluation as a prognostic marker for treatment outcome in advanced breast cancer patients receiving anti-VEGF therapy.