Although the data are too dissimilar to perform a reasonable meta-analysis, the cumulative experience of the discussed immunosuppressive therapies suggests a 20% benefit to treated patients both in terms of improvement in EDSS (26% of treated vs 6% of placebo patients) and sustained progression of disability (19% of treated vs 38% of placebo patients), when outcome analysis is performed at the end of scheduled treatment (see Table 1). In general, the reported beneficial effect of immunosuppression in key trials does not extend much beyond the period of treatment, with the possible exception of autologous bone marrow transplantation. At present, a reasonable strategy to adopt is to consider the use of intensive immunosuppression in those with unequivocally progressive disease in whom a small risk of significant morbidity and mortality can be justified on the basis of poor prognosis. Candidate patients should be at a sufficiently early stage in their disease, and still experience relapses, because in this group immune activation is very likely to be relevant to pathogenesis. Immunosuppressive therapy is unlikely to be of significant benefit to those who already have severe, established neurodegeneration. Additionally, it is obviously more likely to result morbidity and mortality in patients who are wheelchair-bound and /or have very significant bladder involvement. The authors' preference would be pulsed rather than once-off therapy with cyclophosphamide or mitoxantrone, or autologous transplantation. Finally, we believe that immunosuppressive regimens should be administered under appropriately controlled conditions in which patients receive standardised therapy. It is only by adopting this strategy that our knowledge base and hence our ability to assist our patients in making an informed choice will improve.