Background: Attainment of MRD negativity in multiple myeloma (MM) patients is increasingly considered an optimal therapeutic endpoint, but little is known about the MRD evolution in those who achieve this milestone. We investigated the clinical implication of loss of MRD negativity or MRD conversion in patients with ≥VGPR. Methods: We identified and followed 606 patients achieving a sustained ≥VGPR with bone marrow MRD negativity(≥ 2 consecutive reading) following treatment on a total therapy protocol and with a median follow-up of 10 y. All patient had negative PET and MRI DWIBS at enrollment. Serial BM aspirate MRD was determined by 8-color next generation flow (NGF, EuroFlow) with a minimal sensitivity of 10⁻⁵ cells. Results: Most MM patients were considered low risk with a UAMS GEP70 score of ≤ 0.66 (92%; 495/538) . While 60% (364/606) of patients had sustained MRD negativity, the remaining 40% (242/606) experienced MRD conversion with a 5.7 y median time from ASCT and 6.3 y from diagnosis. The risk of clinical relapse was significantly elevated in patients with MRD conversion compared to sustained MRD negativity (73%, 177/242 vs. 5%, 18/364; R.R. = 3.5; P< 0.0001). The median level of MRD positivity (> 0.2 ratio of MM cells to normal plasma cells) also highly correlated with relapse (P< 0.0001). Loss of MRD negativity preceded clinical relapse by a median time of 1.1 years. Loss of MRD negativity without clinical relapse was seen in 27% (65/242). MRD conversion was associated with an inferior PFS and OS (PFS: 10.2 y vs. NR; P< 0.0001, H.R. 18.7; 95% CI 13.3 - 26.3 and OS: 26.1 y vs. NR; P= 0.01, H.R. 1.7; 95% CI 1.1 - 2.6). Furthermore, when MRD conversion was within 5 y of diagnosis compared > 5 y, patients had a worse OS (P< 0.0001, H.R. 17.2; 95% CI 7.8 – 37.8). We also observed that MRD conversion later than 5 years from diagnosis did not affect the OS. In a subset of patients (n = 144) the timing of first MRD negativity following treatment was available. Attainment of MRD negativity within 6 months of diagnosis compared to any time after 6 months was predictive of future MRD conversion (65%, 17/26 vs.42%, 49/118; P = 0.03) and clinical relapse (54%, 14/26 vs.28%, 33/118; P = 0.02). Conclusions: MRD conversion occurs in a significant proportion of MM patients (40%) on long-term follow-up and predicts future clinical relapse. Significance of MRD conversion has a temporal relationship from diagnosis and portray inferior clinical outcome particularly within 5 years of diagnosis.