The Gynecologic Oncology Group conducted a phase II trial of intraperitoneal (ip) cisplatin plus recombinant human alpha-interferon in patients with small-volume persistent/recurrent ovarian cancer. This study was based on the known single agent activity of the drugs administered by the ip route and experimental evidence of cytotoxic synergy between the agents. In 18 evaluable patients, only 1 partial response was observed (5.5% response rate). In an effort to explain these disappointing findings, patients were retrospectively divided into two groups; those with favorable disease (documented response to systemic platinum and absence of surgical findings of diffuse carcinomatosis at laparotomy prior to initiation of ip therapy) or unfavorable disease (no evidence of response to systemic platinum and/or laparotomy findings of diffuse carcinomatosis before ip treatment). The favorable patient population would be predicted to have a far greater chance of responding to local therapy, but only 3 of the evaluable patients fell into this category. In the 15 unfavorable patients, only 1 partial response was observed (7% response rate). We conclude that patients who have failed to demonstrate a response to systemic cisplatin or carboplatin or who have diffuse carcinomatosis at second-look laparotomy are poor candidates for second-line ip cisplatin-based therapy, even if they are considered to have small-volume residual disease (each individual tumor nodules less than 0.5-1 cm). Such patients should be considered for alternative therapeutic strategies.