In order to assess the role of maintenance melphalan and prednisone (MP) in responding multiple myeloma patients, 185 eligible patients who responded to initial MP with stabilization for at least 4 months were randomized to either stop treatment and resume therapy at relapse or to continue MP until relapse. Time to first relapse was significantly shorter in the no maintenance group (P = 0.0011), however 57% of the no maintenance patients had a second response when MP was restarted and others had minor improvement. The time to final progression on MP, which reflects the duration of disease control by MP, was therefore longer for the no maintenance group (median = 39 months) compared to the maintenance group (median = 31 months) although the observed difference was not statistically significant (P = 0.086). Median survival from start of MP in the maintenance group (46 months) was also not significantly different than the no maintenance group (51 months) (P = 0.587). Multifactor analysis of the randomized patients demonstrated shorter total remission duration and shorter survival in patients who had an initially rapid response to therapy or a lesser reduction in serum M-protein concentration.