PURPOSE: To investigate the feasibility of intrapatient dose-escalation methodology for dose-ranging studies of conventional cytotoxics in combination. PATIENTS AND METHODS: Case records were identified for patients with ovarian cancer treated first-line with either single-agent carboplatin or carboplatin and paclitaxel in combination and routinely subjected to a 10% dose escalation in carboplatin at each cycle, towards a target day-22 neutrophil count in the range 1.0-1.5x10(9)/l and a platelet count in the range 75-110x10(9)/l, defining adequate dose. 'Entry level' carboplatin doses were in the range AUC 5.1 to AUC 7.4; paclitaxel was given at 175 mg/m(2) as a 3-h infusion throughout. All drugs were administered three-weekly. RESULTS: The distribution of carboplatin maximum tolerated doses (MTDs) indicated a wide interpatient variation, ranging from AUC 5.4 to AUC 9.8. The median MTD in those receiving carboplatin alone (AUC 6.9) was significantly lower than in those treated with carboplatin and paclitaxel (AUC 7.6) ( P=0.01). Also, paclitaxel had both neutrophil- and platelet-protective effects. CONCLUSIONS: The median MTD documented here using intrapatient dose escalation for carboplatin combined with paclitaxel is remarkably similar to that derived from conventional phase I studies. Furthermore, the striking range of carboplatin MTDs recorded in previously untreated patients may have implications for the wider development of management strategies based on the adequacy of treatment, as defined by the modest levels of dose-limiting toxicity encountered. The ready availability of an expanded set of MTD data by this methodology may also provide more compelling evidence about potential pharmacodynamic drug interactions than may be available from conventional phase I combination studies. These retrospective findings clearly justify further prospective evaluation of intrapatient dose-escalation methodology in dose-ranging studies.