The aim of this study is to test the performance of a matrix model to predict rapid radiological progression (RRP) in a study population of early rheumatoid arthritis (RA) or undifferentiated arthritis (UA) patients. A matrix model using baseline CRP, erosion score, autoantibody status, and initial treatment choice to predict RRP (increase ≥5 points in Sharp-van der Heijde score (SHS) in 1 year) was derived from the BeSt study where patients with active RA (1987-criteria) were treated with initial monotherapy or combination therapy, aiming at low disease activity. In the IMPROVED study, patients with early RA (2010 criteria) and UA were initially treated with methotrexate and prednisone aiming at remission. A receiver operating characteristics (ROC) curve was used to assess the discriminative value of the model to predict damage progression in the IMPROVED population. Four hundred thirty-one out of 479 patients with RA and 106/122 with UA could be categorized as high, intermediate, low, or very low risk for RRP. One patient, with a very low risk profile, showed RRP. Thirty-two other patients (5 %) showed radiological progression ≥0.5 point SHS; none had a high risk profile and 22 had a very low risk profile. The area under the curve (AUC) of the ROC curve was 0.56 (95% CI 0.45; 0.68). A matrix model predicting RRP based on risk factors identified in recent onset active RA according to the 1987-criteria performed poorly in recent onset RA (2010 criteria) and UA. It appears that known risk factors for damage progression lose their impact with early remission steered treatment, so that RRP might be considered a phenomenon of the past.