The objectives of the present study were to determine the following: (a) the maximum tolerated dose (MTD) of melphalan using a 24-h continuous infusion; (b) the clinical toxicity; and (c) the pharmacokinetic characteristics of melphalan at each dose level. Twenty-one patients with refractory solid tumors were enrolled in the study. Melphalan, packaged in 3% sodium chloride, was administered i.v. over a 24-h period. Patients were assigned to one of three escalating dose levels of melphalan: (a) 20 mg/m2 (n = 5); (b) 30 mg/m2 (n = 7); and (c) 40 mg/m2 (n = 6). Each patient underwent pharmacokinetic evaluation during the first cycle of treatment. Melphalan concentrations in plasma were determined by high-performance liquid chromatography. Toxicity was evaluated after each course of chemotherapy. All of the patients were assessable for toxicity and pharmacokinetics, and 20 patients were assessable for response analysis. A total of 50 courses of melphalan was studied. The MTD was 30 mg/m2. The dose-limiting toxicity was neutropenia and thrombocytopenia. Hematotoxicity was reversible (nadir, 14-15 days; recovery, 3.5 and 12.5 days for 30 and 40 mg/m2, respectively), cumulative, and related to the administered dose and to the history of previous therapy. There were six episodes of neutropenic sepsis. Individual pharmacokinetic parameters were estimated using a Bayesian approach and linear elimination kinetics. Data were compatible with a one-compartment model. Relationships have been found between the area under the plasma concentration-time curve and doses and between Css and doses. Moreover, clearance, t1/2 elimination, and volume of distribution did not change statistically with dose, which suggests linear kinetics. Two partial responses were observed in patients with ovarian carcinoma or adenocarcinoma of unknown primary origin, and another patient had stabilization disease. In conclusion, melphalan MTD was determined to be 30 mg/m2 when administered as a 24-h infusion. Hematological toxicity was the dose-limiting toxicity. The most important nonhematological toxicity encountered was nausea and vomiting. The recommended dose for Phase II studies was 30 mg/m2.