The outcome of drug induced pleurodesis has been evaluated in a non-randomised retrospective study of 67 patients with recurrent malignant pleural effusions treated during 1976-83. Fourteen died within 30 days of treatment. Of the remaining 53 patients, 26 studied during 1976-80 were treated with intrapleural mustine hydrochloride, while 27 studied during 1978-83 were treated with intrapleural Corynebacterium parvum. There was no significant difference between the groups in age, sex, or tumour type, most tumours being secondary to bronchial neoplasms. Corynebacterium parvum successfully prevented reaccumulation of the effusion in 24 (92%) patients, while mustine was effective in only 17 (65%) (p less than 0.05). Data on survival were obtained on 47 cases where a single agent was used to induce pleurodesis. In 25 patients treated with C parvum the mean survival was 251 (range 31-1143) days, compared with 119 (range 31-380) days for the 22 patients in whom mustine was used. Survival of the C parvum group was significantly greater (p less than 0.01). The difference was principally due to the greater number of long term survivors in the C parvum group, nine of this group living for more than 180 days (mean 574, range 180-1143 days) compared with only four such survivors in the mustine group (mean 263, range 193-380 days). This study confirms our initial impression that C parvum is highly effective in controlling recurrent malignant pleural effusions. The finding of unexpected long term survivors with C parvum treatment suggests that this treatment may have an additional effect on the progression of disease, a finding that merits further investigation.