This article first reviews the current treatment of lupus nephritis, with a focus on the most serious forms, that is, the proliferative subtypes. Current standards for treatment have been developed empirically. Corticosteroids form the basis of all regimens. Cyclophosphamide given intravenously for prolonged periods is the current gold standard. Azathioprine can be regarded as an effective drug for maintenance treatment of lupus nephritis. Studies on its efficacy in schedules for remission induction are in progress. It has been learned from studies on 'conventional' immunosuppression that randomised, clinical trials should comprise large numbers of patients and a follow-up of many years to elucidate differences between effective strategies. These requirements are not met by any of the 'new' treatments we discuss in this review. There is only limited experience in patients with lupus nephritis with drugs that are currently used for immunosuppression in other autoimmune diseases, such as methotrexate, cyclosporin and high-dose intravenous gammaglobulins, nor with new immunosuppressive drugs that have been developed for immunosuppression in organ transplantation (mycophenolate mofetil, tacrolimus, fludarabine and cladribine). Hormonal therapy with the weak androgen prasterone (dehydroepiandrosterone; DHEA) has no role in treatment of active lupus nephritis. There are interesting experiences with agents that have evolved from progress in immunobiology and in our understanding of immunological processes. These modalities enable more specific immunosuppression and include monoclonal antibodies directed at immune cells, cytokines and components of the complement system, constructs developed to induce tolerance in pathogenic B cells, and gene therapy. Finally, we review data on autologous bone marrow transplantation in patients with systemic lupus erythematosus. We conclude that some strategies (like mycophenolate mofetil) are good candidates for further investigation in large-scale, prospective, randomised trials with prolonged follow-up (which are almost by definition hard to perform). Most new biological agents still are in a pre-clinical phase.