BACKGROUND: Even after complete response (CR) to regional chemotherapy for in-transit melanoma, many patients develop recurrence. Understanding the probability, location, and timing of recurrences can optimize management strategies for this patient population. METHODS: A prospective database identified patients who underwent 81 first-time hyperthermic isolated limb perfusions (HILPs) and 133 first-time isolated limb infusions (ILIs). Response was defined using the response evaluation criteria in solid tumors; recurrence was defined as development of new disease after in-field CR. RESULTS: HILP exhibited a significantly higher CR rate than ILI (44 vs. 28 %, p = .01). Among 36 HILP-CRs and 37 ILI-CRs, the 3-year recurrence rate was 65 % (95 % confidence interval [95 % CI]: 43-79 %) and 85 % (95 % CI: 63-94%), respectively. Median time to first recurrence was longer for HILP-CR than ILI-CR (23 vs. 8 months, p = .02). There was no statistically significant difference in median time to in-field recurrence between HILP-CR and ILI-CR (46 vs. 25 months, p = .15), but HILP-CR showed a longer median time to out-of-field recurrence (42 vs. 14 months, p = .02). Finally, the overall survival (OS) difference between HILP-CR and ILI-CR (3-year survival: 77 vs. 54 %) did not achieve statistical significance (p = .10). CONCLUSIONS: In the largest series comparing patterns of recurrence, we demonstrate that out-of-field recurrence after CR to HILP occurs later than after CR to ILI, though control of in-field disease remains similar. There remains no statistically significant difference in overall survival after CR to the 2 procedures.