BACKGROUND: Talimogene laherparepvec (T-VEC) is the first injectable oncolytic viral therapy approved for in-transit melanoma metastasis, with a reported overall response rate (ORR) of 25% and complete response rate (CRR) of 10%. To ascertain the role of patient selection on outcomes in routine practice, we evaluated the impact of patient, lesion, and treatment factors on clinical response. METHODS: Medical records were extracted for patients with recurrent stage IIIB-IV melanoma completing T-VEC at Duke University Medical Center between 1 January 2016 and 1 September 2018. Kaplan-Meier analysis assessed time to response and survival, while logistic regression measured associations of clinicopathologic status, lesion burden, T-VEC dosing, and use of prior and concurrent therapy with ORR and CRR. RESULTS: Of 27 patients, an objective response was observed in 11 (40.7%), including one patient with partial response (3.7%) and 10 with complete response (37.0%). Time to complete response and overall response was a median 22 weeks (95% confidence interval [CI] 2.0-41.9 weeks and 15.8-28.2 weeks, respectively), and median progression-free survival was 17 weeks (95% CI 0-36 weeks). Logistic regression demonstrated each millimeter increase in maximum lesion diameter predicted decreased ORR (odds ratio [OR] 0.866, 95% CI 0.753-0.995; p = 0.04). Stage IV disease (OR 0.04, 95% CI 0.00-0.74; p = 0.031) and programmed death-1 inhibitor treatment (OR 0.06, 95% CI 0.01-0.74; p = 0.028) also predicted reduced clinical response. CONCLUSIONS: This study corroborates recent data suggesting response rates to T-VEC may be higher than reported in clinical trials, arising in part from patient selection. T-VEC lesion diameter was persistently associated with clinical response and is a readily assessed predictor of successful T-VEC therapy.