Eribulin is one of the newer chemotherapeutic agents approved for use in later line treatment of patients with metastatic breast cancer. Phase III studies have shown a useful clinical response rate for eribulin as well as equivalence to another commonly used drug in metastatic breast cancer, capecitabine. Nevertheless, whether this clinical value observed in trial patients is maintained in patients seen in clinical oncology practice, outside trials remains a question. Several series published over the last few years sought to answer this question. The current paper carries out a pooled analysis of these retrospective series to obtain efficacy and toxicity data for eribulin in metastatic breast cancer patients treated off trials. Thirteen series with a total of 1095 patients were identified. Pooled estimates of response rate and clinical benefit rate were 20.1% (95% confidence interval: 16.3–23.9%) and 46.3% (95% confidence interval: 39.4–53.2%) respectively. These were somewhat higher than the response rate and clinical benefit rate observed in a pooled analysis of two randomized phase III trials (14.9 and 30.9%, respectively, conducted in an intension-to-treat manner). In contrast overall survival was longer in the phase III trials (median 15.2 months) than in the retrospective studies (pooled estimate 9.8 months). All grades toxicities were similar in practice compared with trials with slightly higher grade 3 toxicities (46.1 vs. 38.7%) but lower grade 4 toxicities (17.2 vs. 27.7%) in patients off trials.