The three dosimetric schedules currently used in isolated limb perfusion with melphalan for malignant melanoma of the lower limb were compared in a series of 51 patients. The doses prescribed by each of the three methods (based on total body weight (TBW), limb tissue volume (LTV) and total blood volume in the perfusion circuit (TBV)) were calculated for all patients and were then compared using Wilcoxon's signed-rank test. This revealed that the method based on TBV consistently prescribed much lower doses of drug than either of the other two methods. Pharmacokinetic profiles of melphalan obtained by HPLC analysis of blood samples during the procedure also showed that the method did not reliably predict the concentration of melphalan achieved in the perfused limb. The dosimetric schedule based on LTV prescribed slightly higher doses than that based on TBW. However, the technique is more difficult to practise due to the problems of measuring the limb volume by immersion. We conclude that the dosimetric schedule based on TBW is the most appropriate by virtue of its simplicity, the high doses of melphalan which it prescribes, and the well-controlled toxicity which it produces.