To determine the optimum dose of OK-432 for intrathoracic administration, a multicenter randomized phase II trial was conducted in patients with malignant pleural effusion due to non-small cell lung cancer. Patients with histologically- or cytologically-proven malignant pleural effusions were randomized to arm A (10 Klinische Einheit (KE) of OK-432) or arm B (1 KE of OK-432). OK-432 was injected intrapleurally over 30 min on days 1 and 3 and the chest tube was clamped for 6 h. If control was inadequate on day 8, 10 KE was administered on days 8 and 10 in each treatment arm. Forty patients were enrolled and 38 patients were eligible (19 in arm A and 19 in arm B). The effusion control rate on day 8 was 79% in arm A and 53% in arm B, while control rates on day 28 were 74% and 84%, respectively. The median drainage time after administration was significantly shorter in arm A (4.0 +/- 1.2 days) than in arm B (7.0 +/- 1.7 days). The total drainage volume was also significantly less in arm A than in arm B. No grade 4 toxicities or treatment-related deaths were observed in either treatment arm. Intrathoracic injection of OK-432 is a feasible treatment for malignant pleural effusion. Although the malignant pleural effusion control rate was equivalent in each treatment arm, faster control and less drainage were achieved in arm A. A dose of OK-432 10 KE/body is, therefore, recommended for further trial.