Thirty dogs with spontaneously occurring malignant neoplasms were treated monthly with carboplatin (CBDCA) given as a 30-minute intravenous infusion in a dose escalation study. Twenty-eight dogs were considered evaluable for toxicity. The maximally tolerated dose of CBDCA was conceptually defined as that dose, determined by logistic regression analyses of toxicity data, resulting in a 50% incidence of moderate toxicity (MOD50) or a 5% incidence of severe toxicity (SEV5). Each designated maximally tolerated dose was calculated for the first course of treatment only and for the first and second courses of treatment combined to estimate cumulative drug toxicity. The MOD50 and SEV5 for the first treatment course were 340 and 278 mg/M2, respectively. MOD50 and SEV5 values for the first plus second treatment courses were 327 and 231 mg/M2, respectively. The nadir of neutrophil and platelet counts occurred approximately 14 days after treatment. The mean neutrophil and platelet values for all dogs experiencing myelosuppression during the first two treatment courses were 1541/μL and 62,600/μL, respectively. Nonparametric pharmacokinetic analysis of plasma CBDCA values suggested that half-life (T1/2), area-under-the-curve and total body clearance (CLb) were not dose dependent. Volume of distribution (VDss) significantly increased with dose only between 100 and 150 mg/M2, not between 150 and 300 mg/M2. Dose-independent serum CBDCA pharmacokinetic disposition indicates that detailed investigation of tissue CBDCA distribution would be warranted and may identify novel dosing strategies that could improve the therapeutic index of CBDCA by minimizing toxicity. (Journal of Veterinary Internal Medicine 1993; 7:235–240. Copyright © 1993 by the American College of Veterinary Internal Medicine.)