Published, randomized trials of high-versus-low-dose steroids in renal transplantation, individually and collectively, are too small (median trial size: 72 patients) to reach authoritative conclusions about mortality, graft failure, transplant failure, and the incidence of certain complications. The implications of this review for future therapeutic trials in renal transplantation are that randomization of 400 patients should be seen as minimum Phase III trial size, and that a research strategy is needed to encourage effective multicenter collaboration. One such strategy is described beginning with a single-center randomized Phase II trial to compare times to first rejection episode, thus rendering prior evidence on which the case for multicenter collaboration, effectively in an established trial, can be judged. Analysis by intention to treat and consistent reporting of the number of patients who experience certain classes of complication are two of four ground rules on which we assessed trials when combining evidence from similar studies. The other ground rules are randomization and common treatment theme. The statistical method of formally pooling results is outlined.