OBJECTIVES: To identify predictive factors for lymphoproliferative disorders (LPDs) that persist after methotrexate (MTX) withdrawal (Persistent-LPD) and the optimal treatment for rheumatoid arthritis (RA) after LPD regression. METHODS: Among 3666 patients with RA treated with MTX in our department from 2006 to 2017, 26 cases of LPD that regressed after MTX withdrawal (Regressive-LPD) and 25 cases of Persistent-LPD were compared. Multivariate logistic analysis was performed to identify predictive factors for Persistent-LPD. Retention rates of biological disease-modifying antirheumatic drugs (bDMARDs) were calculated using the Kaplan-Meier Method. RESULTS: In Persistent-LPD, the incidence of diffuse large B-cell lymphoma was higher (76%). The overall 2-year survival rate was 83.9%: 95.8% for Regressive-LPD and 71.0% for Persistent-LPD. The International Prognostic Index (IPI) risk classification was useful for predicting Persistent-LPD. bDMARDs were introduced in 38 RA patients after LPD regression. Unadjusted retention rate of bDMARDs in the 51 LPD patients was significantly lower than that in the 1668 non-LPD RA patients in our bDMARD cohort (controls) (p = 0.029). The 1-year retention rates for bDMARDs were 69% and 64% for tocilizumab and abatacept, respectively vs. 46% for TNF-inhibitor (TNFi). CONCLUSION: Risk assessment using IPI predicted Persistent-LPD. After LPD regression, non-TNFi tended to have higher retention rates.