<h3>Objectives:</h3> Eribulin (E), a Halichondrin B analog from the marine sponge H. okadai has clinical efficacy in pretreated metastatic breast cancer and liposarcoma as well as preclinical antitumor activity in squamous cell carcinoma (SCC). We conducted a 2-stage Phase 2 study of E in patients (pts) with advanced/recurrent cervical cancer (CC) to examine its clinical activity and evaluate potential biomarkers for predictors of response. <h3>Methods:</h3> Pts with advanced/recurrent CC after ≤1 prior chemotherapy (CT) regimen, measurable disease and ECOG performance status ≤2 were treated with E (1.4mg/m<sup>2</sup> IV day 1 and 8, every 21 days) with tumor assessments every 2 cycles. Primary endpoint was 6-month progression-free survival PFS6; secondary were best overall response (RECISTv1.1), toxicity (CTCAEv4.03) and overall survival (OS); and exploratory were associations of tumor biomarkers involved in apoptosis/proliferation, the unfolded protein response and tubulin subtypes with clinical activity. A total of 30 evaluable pts would ensure 80% power when the true PFS6=26% with a 1-sided α ≤0.1 (Ho: PFS6=10%). A prespecified futility analysis gating stage 2 was set if 0/15pts showed at least SD at 6 months (mos). Immunohistochemistry was performed on archival tumor samples. Overexpression was defined when both intensity and distribution scores are ≥2. <h3>Results:</h3> A total of 32 pts were enrolled, median age 51 years (range 29-76), 22 had SCC, the median number of cycles was 4 (range 1-77). A total of 29 pts had received prior CT with cisplatin/gemcitabine (12) and cisplatin/paclitaxel/bevacizumab (12) as the most common regimens. A total of 14 pts had received prior paclitaxel (PTX). A total of 6/32 pts (19%) achieved > PFS6; median PFS is 2.5 mos (95% CI: 1.2, 4.2). Median OS is 6.5 mos (95% CI: 4.5, 12.7). A total of 2 pts were inevaluable for response having received less than 2 cycles. Among the 30 evaluable pts, 1 (3%) had a complete response, 5 (17%) had a partial response and 13 (43%) had stable disease (clinical benefit rate 63%). Grade 3/4 adverse events occurring in > 10 % of pts are anemia (12pts), neutropenia (7pts) and leukopenia (6pts). A total of 2 pts were removed from study due to paresthesia. Pts who received prior PTX responded less favorably than those who did not (p=.002) and had a significantly shorter PFS at 1.2 mos (95% CI: 1.1, 2.0) compared to 3.5 mos (95% CI: 2.6, 5.4) for those who did not (p=.008). Analysis of correlative predictors of response revealed significant associations between βII and βIII tubulin subtypes as well as BAX. Patients who did not overexpress βII and BAX were more likely to respond to E (p=.040, p=.002, respectively). On univariate analysis, PFS was shorter in patients with BAX overexpression 1.2 mos (95% CI: 1.1, 2.6) vs 4.6 mos (95% CI: 2.7, 7.8, p=.006) in those without. This was also observed in OS with 4.3 mos (95% CI: 2.7, 6.5) in those who overexpressed BAX vs 1.3 mos (7.7, 28.3, p=.006) in those who did not overexpress. In multivariate analyses, prior PTX (HR 4.80, 95% CI: 1.12, 20.7), βIII tubulin overexpression (HR 5.22, 95% CI: 1.52, 17.93), and BAX overexpression (HR 3.61, 95% CI 1.11, 11.73) remained significant factors in PFS. Similar results were observed for OS. <h3>Conclusions:</h3> Eribulin shows some limited activity in patients with recurrent/advanced CC with a favorable toxicity profile. Prior paclitaxel exposure is associated with decreased E response. βII, βIII tubulin subtypes and BAX are predictors of response and survival. While the primary end point was not met, Eribulin may be an option for women with paclitaxel-naïve recurrent/advanced CC.