Hypoxic antiblastic stop-flow perfusion (SFP) has recently been proposed as a therapeutic option for patients with locally advanced tumors. We report on the clinical and pharmacological results of our prospective study of limb SFP for the treatment of in transit melanoma metastases. Twenty-three patients with limb-sited melanoma metastases were treated with melphalan (10 mg/l) based pelvic (n=11, group A) or femoral (n=12, group B) SFP under hypoxic conditions. Systemic and locoregional toxicity, tumor response rate, and local progression-free survival were analyzed. Melphalan concentrations were measured in the perfusate and systemic circulation during SFP, and after 30-min hemofiltration. Perfusate-to-plasma leakage was assessed using a scintigraphic method. No postoperative deaths occurred. Mild locoregional toxicity was observed in 5 patients (18%), and systemic toxicity was mild to severe in 8 patients (30%), the incidence being higher in group A. Tumor response rate (complete + partial response) and time to local disease progression were significantly different in group A and B (9% vs 58% and 7 vs 13 months, respectively). The pharmacokinetic study showed that pelvic SFP was associated with a higher leakage rate and a lower area under the curve ratio than femoral SFP (44% vs 31% and 5.6 vs 9.8, respectively). Limb SFP is a feasible and relatively simple procedure. Toxicity and tumor response rates strictly depend upon drug leakage control. Further efforts should be made to exploit the potential anti-tumor activity of this novel locoregional drug delivery system.