Results after an average follow-up of 3 years are presented on 485 patients in the 3rd MRC therapeutic trial in myelomatosis. The 353 non-azotaemic patients (199 now dead) were randomized between i.v. cyclophosphamide (CY) and oral melphalan with prednisone (M & P). THose treated with M & P fared slightly, but non-significantly, better. The 132 azotaemic patients (111 now dead) were randomized between i.v. CY and a 4-drug regimen, and both groups fared equally badly. Finally, after one year of the allocated treatment, 297 survivors (126 now dead) were randomized either to stop all treatment until evidence of relapse was obtained, or to continue treatment with azathioprine and vincristine, interrupted every 3 months for a course of the first-allocated treatment. The overall results suggested that maintenance therapy was beneficial, though the results were not statistically significant. Most of the difference was found among the few patients with unfavourable prognostic features who survived one year and were eligible for this randomization. In this, as in the two previous MRC trials, no striking differences have emerged between the therapeutic effects of different schedules of melphalan and/or CY. Consequently, a regimen of intermittent oral melphalan (with or without prednisone) seems satisfactory, because it is among the least toxic and most convenient. The 4th myeloma trial, now beginning, seeks to discover whether the addition of vincristine to the regimen can improve these results.