Background Renal impairment (RI) is a common complication of multiple myeloma (MM) affecting approximately 20%-30% of the patients at diagnosis. The presence of severe RI is associated with an increased risk of early death and poor prognosis in myeloma. The impact of autologous stem cell transplant (ASCT) on outcome in severe RI patients remains controversial. Therefore, we performed a retrospective real-world analysis to clarify the effect of ASCT on the survival of MM patients with severe RI. Methods Clinical data of 337 MM patients with severe RI at diagnosis were collected from three centers in China from April 2004 to January 2020. Severe RI was defined as an increase creatinine >177 umol/L (or 2mg/L) or an estimated glomerular filtration rate ≤40 ml/min/1.73 m2 using the simplified modification of diet in renal disease (MDRD) formula. RI caused by other confirmed pathological diagnosis except for tubular nephropathy was excluded. We excluded patients underwent salvage ASCT (6 cases), patients older than 65 years (107 cases), and traditional introduction therapy (2 cases). At last, there were 232 cases for analysis, containing 59 cases received ASCT. All enrolled patients were treated with novel-agent introduction. Patients were divided into the ASCT group and the non-ASCT group. Hematological and renal response were assessed based on the International Myeloma Working Group (IMWG) criteria. Results The median age for all patients at diagnosis was 55 years old (range: 37-65 years old). 37% presented with free light chain myeloma. There were 4 patients underwent double-ASCT in the ASCT group. All dialysis patients in ASCT group were off dialysis after induction therapy. Baseline profiles between ASCT and non-ASCT group were matched in sex, M-protein isotype, ISS stage, and induction therapy. More patients were on dialysis at diagnosis in non-ASCT group (34.7% vs 19.3%, P=0.04). There were 59/159 patients presented unfavorable cytogenetic abnormality (del(17p), t(4;14), t(14;16)), including 40 cases (34.2%) in non-ASCT group and 19 cases (41.3%) in ASCT group (P=0.47). The 1q amplification presented 21.4% in non-ASCT group and 22.2% in ASCT group (P=0.56). In terms of hematological response, at least VGPR accounts for 88.1% in the ASCT group and 50.3% in the non-ASCT group (P<0.001). The overall renal response was significantly higher in the ASCT group than that in the non-ASCT group (94.9% vs 54.3%, P<0.001). There was the similar tendency for at least renal PR between ASCT and non-ASCT groups (88.1% vs 41.3%, P<0.001). The transplant-related mortality (within 3 months after ASCT) was 6.8%. Early mortality within 1 year was 6.7% in the ASCT group and 19.8% in the non-ASCT group (P=0.002). The median follow-up for all patients was 37.6 months (range: 1-157 months) and the median overall survival (OS) was 49.5 months (95% CI:42-57 months). In the ASCT group, the median OS was much longer than ASCT group (57.3 months vs 42.8 months, P =0.052). But there was no difference in PFS between the two groups (36.9 months vs 26.5 , p=0.348). As the log-rank test didn't hold the proportional hazards assumption, it would have low power. We use the two-stage landmark analysis to further determine the effect of ASCT on survival. In landmark analysis, 3-year PFS rate in ASCT group was significantly higher (54.0% vs 41.9%, P=0.024). Within 5-year follow-up, OS was significantly longer in the ASCT group than that in non-ASCT group (OR=0.61, 95%CI: 0.39-0.97, P=0.04). Beyond the 5-year time point, OS was similar between the two groups (OR=1.59, 95%CI: 0.60-4.20, P=0.35). In univariate analysis, at least renal MR, at least hematological VGPR, PLT>100×109/L, and BMPC ≥60% were correlated with longer OS. In multivariate analysis, at least renal MR (HR=0.61, 95%CI: 0.39-0.96, P=0.031), at least hematological VGPR (HR=0.41, 95%CI: 0.27-0.63, P<0.001), and PLT>100×109/L (HR=0.43, 95%CI: 0.28-0.66, P<0.001) were independent predictor of OS. Although upfront ASCT could remarkably prolong OS in severe RI myeloma, it was not confirmed in multivariate analysis. Conclusions In the novel agent era, role of upfront ASCT in myeloma with severe RI remains uncertain. In our analysis, it seems the benefit of the OS from novel-drug induction and upfront ASCT was observed in the first 5-year, but the long-term survival benefit was limited. Instead, deep hematological response and renal recovery was strong predictors for OS.