A disease-oriented approach to the discovery of novel platinum anticancer drugs has been established through the setting up of parallel human ovarian-carcinoma cell lines and xenografts. The correlation between in vitro and in vivo antitumour activity was determined for four reference platinum agents (cisplatin, carboplatin, iproplatin and tetraplatin) in eight companion lines. Two methods of assessing antitumour effect were used in vitro (tritiated thymidine incorporation and sulforhodamine B staining) and three were applied in vivo [28-day treated/control (T/C) ratio, growth delay and specific growth delay]. In vitro, large differences in cytotoxicity across the cell lines were observed for each drug. This was also reflected in the xenografts for cisplatin and carboplatin and, to a lesser extent, for iproplatin. A correlation analysis of in vitro vs in vivo data revealed a high, statistically significant positive correlation for cisplatin and a strong positive correlation for carboplatin. However, for the two platinum(IV) drugs, the correlation was less good. In particular, tetraplatin was markedly less active in vivo (showing a general lack of activity against all of the tumour lines) than its in vitro potency against the cell lines predicted, resulting in poor correlation coefficients. These human tumour panels may be valuable for the elucidation of both cellular/molecular and corresponding in vivo pharmacological mechanisms of platinum drug resistance. Moreover, the HX/62 and SKOV-3 tumour lines, which exhibit a level of intrinsic resistance to the four reference agents both in vitro and in vivo (and which were derived from patients who had not received prior platinum therapy), represent particularly useful evaluation models for the discovery of novel broad-spectrum platinum drugs.