Abstract Background and hypothesis Primary membranous nephropathy (PMN) is usually caused by anti-PLA2R autoantibodies. There are different therapeutic options according to baseline risk. Novel biomarkers are needed to optimize risk stratification, and predict and monitor the response to therapy, as proteinuria responses may be delayed. We hypothesized that plasma or urinary cytokines may provide insight into the course and response to therapy in PMN. Methods Overall, 192 data-points from 34 participants in the STARMEN trial, randomized to tacrolimus-rituximab (TAC-RTX) or corticosteroids-cyclophosphamide (GC-CYC), were analyzed for plasma and urine cytokines using a highly sensitive chemiluminescence immunoassay providing a high-throughput multiplex analysis. Results Baseline (pretreatment) urinary CXCL13 predicted the therapeutic response to TAC-RTX. Cytokine levels evolved over the course of therapy. The levels of nine plasma and six urinary cytokines correlated with analytical parameters of kidney damage and disease activity, such as proteinuria, eGFR and circulating anti-PLA2R levels. The correlation with these parameters was most consistent for plasma and urinary GDF15, plasma TNF-α and urinary TWEAK. Decreasing plasma GDF15 levels were associated to response to GC-CYC. Four clusters of cytokines were associated to different stages of response to therapy in the full cohort, with the less inflammatory cluster associated to remission. Conclusion In conclusion, PMN displayed characteristic plasma and urine cytokine patterns that evolved over time as patients responded to therapy. Baseline urinary CXCL13 concentration could be a prognostic marker of response to TAC-RTX.