Valuable information was collected on the medical history and clinical course of 1273 patients entered in clinical trials with Adriamycin (ADR) carried out in 12 European cancer centers. A coded patient form was used for the data collection carried out in each center by a qualified physician following a guideline which was discussed and accepted by all of the participants. The aim of the study was to define the incidence, characteristics, and possible co-factors of the cardiomyopathy (CMP) in patients treated with combination chemotherapy regimens including ADR. The mean total dose of ADR was 268 mg/m2 (range, 15--1251 mg/m2), and 5.1% of the patients received a total dose of greater than 550 mg/m2. A "definite" ADR-related CMP was observed in 1.7% of the cases; another 3% of the cases were reported as "possible" ADR-CMP since the role played by the drug could not be clearly defined. "Definite" ADR-CMP was fatal in eight patients (0.6%) while "possible" ADR-CMP was fatal in 13 patients (1.0%). Among the possible co-factors examined, the following ones were found to be significantly associated with the occurrence of a "definite" ADR-CMP: (a) total dose of ADR; (b) vincristine when given both before and concomitantly with ADR; (c) bleomycin when given before ADR; and (d) radiotherapy to the mediastinum when given concomitantly with ADR. Furthermore, none of 182 patients receiving ADR by slow infusion developed a "definite" ADR-CMP, while 2% of the patients treated by bolus injection did so. The occurrence of a "possible" ADR-CMP was found to be significantly associated with two pre-existing pathologic conditions (electrocardiogram [ECG] abnormalities and hypertension) but not with the treatment-related co-factors for the "definite" ADR-CMP mentioned above. Other variables examined, such as sex, age, cancer type, baseline liver function, and cyclophosphamide treatment, did not seem to influence the risk of ADR-CMP. Data on ECG changes occurring during ADR treatment were also reported and their incidence was found to be strictly related to the frequency of the ECG monitoring.