Abstract Objectives In patients with LN, clinical and histological responses to treatment have been shown to be discordant. We investigated whether per-protocol repeat kidney biopsies are predictive of LN relapses and long-term renal function impairment. Methods Forty-two patients with incident biopsy-proven active proliferative (class III/IV±V) LN from the database of the UCLouvain were included in this retrospective study. Per-protocol repeat biopsies were performed after a median [interquartile range (IQR)] time of 24.3 (21.3–26.2) months. The National Institutes of Health activity index (AI) and chronicity index (CI) scores were assessed in all biopsies. Results Despite a moderate correlation between urinary protein/creatinine ratios (UPCR) and AI scores at repeat biopsy (r = 0.48; P = 0.001), 10 patients (23.8%) with UPCR < 1.0 g/g still had a high degree of histological activity (AI > 3). High AI scores (continuous) in repeat biopsies were associated with an increased probability and/or shorter time to renal relapse (n = 11) following the repeat biopsy [hazard ratio (HR) = 1.2, 95% CI: 1.1, 1.3; P = 0.007], independently of proteinuria levels. High CI scores (continuous) in repeat biopsies were associated with a sustained increase in serum creatinine levels corresponding to ≥120% of the baseline value (HR = 1.8, 95% CI: 1.1, 2.9; P = 0.016) through a median (IQR) follow-up time of 131.5 (73.8–178.2) months, being also the case for acute tubulointerstitial inflammation and interstitial fibrosis/tubular atrophy in repeat but not baseline biopsies. Conclusion Our results highlight the usefulness of per-protocol repeat biopsies, herein performed after a median time of 24 months from baseline, as an integral part of the treatment evaluation, also in patients showing adequate clinical response.