The pharmacokinetics of carboplatin, ultrafilterable platinum, and total platinum after intraperitoneal (i.p.) administration were studied in peritoneal fluid, plasma, red blood cells (RBCs), and urine during a phase-I trial in patients with minimal, residual ovarian cancer. Samples were collected from 7 patients who had received carboplatin (200-500 mg/m2) in 21 dialysis fluid. The fluid was withdrawn after a 4-h dwell. Platinum concentrations were measured by flameless atomic absorption spectrometry, and intact carboplatin was determined by HPLC with electrochemical detection. Peak concentrations of carboplatin in plasma were obtained 2 h after the end of instillation. The mean ratio of peak concentrations of carboplatin in instilled fluid and plasma was 24 +/- 11. The peritoneal clearance of carboplatin was 8 +/- 3 ml/min, which was 12 times less than the plasma clearance (93 +/- 32 ml/min). Due to this clearance ratio, the AUCs for the peritoneal cavity were about 10 times higher than those for plasma. On average, 34% +/- 14% of the dose was still present in the instillation fluid that had been withdrawn after a dwell time of 4 h. In plasma, the mean value of AUC/Dnet (Dnet = Dose - amount recovered from the peritoneal cavity) after i.p. administration was comparable with that of AUC/D after i.v. administration. This means that unrecovered carboplatin (66%) was completely absorbed from the peritoneal cavity. It may be expected from this bioavailability that the maximum tolerated dose (MTD) of i.p.-administered carboplatin with a 4-h dwell is around 1.5 times higher than that after i.v. administration. Overall pharmacokinetic parameters of carboplatin and platinum in plasma were comparable after i.p. and i.v. administration.