Fifty dogs with refractory or disseminated spontaneous tumours were evaluated in two independent phase I studies using either carboplatin (CBDCA) alone or CBDCA plus whole body hyperthermia (WBH). CBDCA was administered as a 30 min infusion at the onset of the plateau phase of WBH in dogs receiving combined treatment. Serum samples were collected and drug disposition was determined in both treatment groups. The dose-effect relationship was mathematically described with a logistic regression model developed from categorical toxicity data accumulated throughout the first two treatment courses in all dogs. The maximum tolerated dose (MTD) was defined as that dose which resulted in a 50% probability of achieving moderate or severe toxicity. The only toxicities observed were neutropenia and thrombocytopenia, which were dose-dependent. The nadir occurred between 7 and 14 days following treatment. A significant decrease in the area under the serum CBDCA versus time curve for dogs undergoing WBH was consistent with increased tissue binding of the drug as well as increased urinary eliminations. Serum AUC values determined following the first course of treatment were predictive of subsequent toxicity in both treatment groups. The MTD (95%CI) for CBDCA and CBDCA/WBH were estimated to be 318(44) and 239(51) mg/M2 respectively (p = 0.08). A randomized phase II evaluation should be initiated to determine if a therapeutic gain can be achieved using combined CBDCA and WBH. Further refinement of the CBDCA dose in such a trial should be based on both pharmacokinetic parameters and normal tissue response.