Purpose: To determine the maximally tolerated dose, toxicity, and pharmacokinetics of i.p.gemcitabine. Experimental Design: Patients had peritoneal carcinomatosis. Gemcitabine (40, 80, 120, or 160 mg/m 2 ) was administered into the peritoneal cavity in 2 L of warmed saline on days 1, 4, 8, and 12 of a 28-day cycle. Results:Thirty patients received 63 (median, 2; range, 0-6) courses.Tumors included ovary (14), uterus (2), colon (6), pancreas (3), and others (5). Dose-limiting toxicity included nausea, vomiting, diarrhea, dyspnea, fatal respiratory failure, and grade 3 elevation of alanine aminotransferase in three patients. Hematologic toxicity and pain were Vgrade 2. Three patients had decreased or resolved ascites. Of 19 patients evaluable for response, 10 had stable disease (median, 3.5 courses) and 9 had progressive disease. The median peak peritoneal concentration was1,116-fold (range, 456-1,886) higher than the peak plasma level. Plasma and peritoneal levels were undetectable within 8 to 12 h. At 120 mg/m 2 , the median peritoneal area under the concentration versus time curve (AUC) was 82,612 ng/mL Â h (range, 53,296-199,830) and the plasma AUC was 231 ng/mL Â h (range, 47.6-259.5). The mean peritoneal advantage (AUC peritoneal / AUC plasma ) was 847 (range, 356-1,385).