BackgroundRelapse after initial remission reduces renal survival in patients with primary membranous nephropathy (PMN). In this study, we aim to identify risk factors of relapse in PMN and construct a model to identify patients at high risk of relapse early.MethodsWe conducted a multi-center retrospective study using the China Renal Data System database, which includes data from 24 urban academic centers across China. A prediction model based on the Cox proportional hazards model was derived in the derivation group and validated in the validation group.Result515 patients with biopsy-proven PMN achieving initial remission were enrolled. 32.62% of patients subsequently relapsed during a median of 6.08 months. Lower serum albumin (Alb) (per 1 g/L decrease, hazard ratio [HR] =1.48, 95% confidence interval [CI] 1.29–1.78, p < 0.001), lower estimated glomerular filtration rate (eGFR) (per 10 mL/min/1.73m2 decrease, HR =1.14, 95% CI 0.97–1.49, p < 0.001), higher serum complement 4 (C4) (per 0.1 g/L increase, HR =1.89, 95% CI 1.32–3.22, p = 0.012), partial remission (PR) (HR =2.28, 95%CI 1.74–4.04, p < 0.001), and treatment with calcineurin inhibitors (CINs) (HR =1.33, 95%CI 1.04–1.64, p < 0.001) at the time of remission were risk factors for relapse. C-statistic, time-dependent areas under the receiver operating characteristic curve, and calibration plots confirmed that the model had excellent discrimination and calibration in predicting PMN relapse. The anti-phospholipase A2 receptor antibody (aPLA2Rab) titers and pathologic features did not substantially improve the model.ConclusionOur study confirms the well-known low Alb and eGFR, PR, and treatment of CNIs at the time of remission as risk factors for PMN relapse, but aPLA2Rab and pathologic features may not predict relapse. In addition, it is the first study to show serum C4 is associated with PMN relapse. We suggest that complement-targeted therapies may be a potential therapy to prevent PMN relapse.