Uveal melanoma (UM) is a rare form of melanoma without effective therapy. The biology of UM relies on several heat-shock protein 90 (Hsp90)-dependent molecules such as MET, MEK and AKT, making Hsp90 inhibition a rational approach. Patients with stage IV UM, measurable disease, and no previous chemotherapy were eligible. Patients received either ganetespib 200 mg weekly (cohort A) or 150 mg twice a week (cohort B). Primary endpoint response rate (RR) was assessed by RECIST. A total of 17 patients were accrued for this study, with seven in cohort A and 10 in cohort B. Liver metastases were present in 59%. Response outcomes included one partial response, four stable disease, 11 progressive disease, and one withdrawal for ORR: 5.9% and disease control rate of 29.4%. Progression-free survival was 1.6 months (cohort A) and 1.8 months (cohort B). Overall survival was 8.5 months (cohort A) and 4.9 months (cohort B). An overall 31% of adverse events were grade 3-4 and were mostly related to gastrointestinal toxicities. Early on-treatment (1 months) positron emission tomography showed reduction in metabolic activity in 24% of patients, suggesting a pharmacodynamic effect of Hsp90 inhibition. These early metabolic changes did not seem to be durable and/or clinically significant in relation to the 2-month response assessment. Hsp90 inhibition with ganetespib resulted in modest clinical benefit on two dosing schedules and was associated with significant, although manageable, gastrointestinal toxicity. Evidence of pharmacodynamic activity for Hsp90 inhibition was observed via positron emission tomography, which did not translate into clinical benefit, suggesting rapid development of resistance.