HYPOTHESIS: Recent evidence indicates that tumor response rates after isolated limb perfusion (ILP) are improved when tumor necrosis factor (TNF) is added to the locoregional perfusion of high doses of chemotherapy. Other factors, related to the patient or the ILP procedure, may interfere with the specific role of TNF in the early hemodynamic response after ILP with TNF and high-dose chemotherapy. DESIGN: Case-control study. SETTING: Tertiary care university hospital. PATIENTS: Thirty-eight patients with a locoregionally advanced tumor of a limb treated by ILP with TNF and high-dose chemotherapy (TNF group) were compared with 31 similar patients treated by ILP with high-dose chemotherapy alone (non-TNF group). INTERVENTIONS: Swan-Ganz catheter hemodynamic recordings, patients' treatment data collection, and TNF and interleukin 6 plasma level measurements at regular intervals during the first 36 hours following ILP. MAIN OUTCOME MEASURES: Hemodynamic profile and total fluid and catecholamine administration. RESULTS: In the TNF group, significant changes were observed (P<.006): the mean arterial pressure and the systemic vascular resistance index decreased, and the temperature, heart rate, and cardiac index increased. These hemodynamic alterations started when the ILP tourniquet was released (ie, when or shortly after the systemic TNF levels were the highest). The minimal mean arterial pressure, the minimal systemic vascular resistance index, the maximal cardiac index, the intensive care unit stay, and the interleukin 6 maximal systemic levels were significantly (P<.001 for all) correlated to the log(10) of the systemic TNF level. In the non-TNF group, only a brief decrease in the blood pressure following tourniquet release and an increase in the temperature and in the heart rate were statistically significant (P<.006). Despite significantly more fluid and catecholamine administration in the TNF group, the mean arterial pressure and the systemic vascular resistance index were significantly (P<.001) lower than in the non-TNF group. CONCLUSIONS: Release of the tourniquet induces a blood pressure decrease that lasts less than 1 hour in the absence of TNF and that is distinct from the septic shock-like hemodynamic profile following TNF administration. The systemic TNF levels are correlated to this hemodynamic response, which can be observed even at low TNF levels.