Landmark analyses are used to investigate the importance for survival of achieving complete response (CR), an important initial goal of myeloma therapy. With median times to CR in Total Therapy (TT) trials of approximately 1 year, this approach excludes a sizeable fraction of patients dying before such a landmark. To permit inclusion of all trial participants, we investigated the prognostic implications of both onset and duration of CR as time-dependent variables. Superseding the adverse effects of cytogenetic abnormalities and other standard prognostic parameters, both failure to achieve CR (non-CR) and, especially, loss of CR (los-CR) were independently associated with inferior survival in TT1, TT2, and TT3 protocols. In the context of gene array-defined risk, available in TT2 and TT3 subsets, both los-CR and non-CR terms were retained in the survival model as dominant adverse variables, stressing the prognostic importance of sustaining CR status, especially in high-risk disease.