The data we review indicate that in adults with ALL in first remission intensive chemotherapy and radiation given before a transplant is more effective in eradicating leukaemia than current chemotherapy. This is not so in adults with AML in first remission where more intensive therapy does not reduce the likelihood of relapse. Furthermore, leukaemia relapse because of re-infused leukaemia cells is an important issue in autotransplants for ALL. Whether re-infusing leukaemia cells would be important in AML were more effective pretransplant therapy developed is unknown. Presently, there appears to be little sense to test in vitro approaches to remove leukaemia cells in autotransplants for AML because efficacy cannot be evaluated. (A randomized trial is an exception but would not be expected to show a difference.) Another conclusion is that attempts to induce GVHD in autotransplant recipients, such as by using cyclosporine post-transplant (Jones et al, 1989), are more likely to succeed in AML than ALL since the impact of GVHD is substantially greater (Horowitz et al, 1990). However, the GVHD-related antileukaemia effect in AML is associated with chronic GVHD whereas cyclosporine treatment of autotransplant recipients results in acute GVHD. Also, attempts to separate clinical and antileukaemia effects of GVHD were unsuccessful (Sullivan et al, 1989). Another caution is that in AML we detected an immune antileukaemia effect distinct from GVHD (termed GVL) only after HLA-identical sibling transplants. Since GVL was absent in twins it is unlikely to operate after autotransplants. In summary, there is sense in studying autotransplants in adults with acute leukaemia in first remission. However, there are currently no convincing data that autotransplants are superior to current therapy. More intensive treatment seems effective in ALL; the focus should be on increasing the antileukaemia efficacy of pretransplant therapy and on attempts to remove leukaemia cells from the graft. In AML, there is no evidence that more intensive therapy is more effective. This problem needs resolution before evaluating attempts to remove leukaemia cells from the graft. The best place to test new pretransplant regimens is in twins and recipients of HLA-identical sibling transplants without GVHD. The data and ideas we review and discuss should be useful in planning clinical trials.