Special problems may arise when the principles of the randomized trial are used to compare operative methods. In the trial design that is used at present, each surgeon performs the operatons to be compared in a random order. A prerequisite for a trial so designed to result in a fair comparison is that the participating surgeons are equally conversant with both techniques. If they are not, the design may imply systematic bias in favor of operations that are in wide use and bias in favor of technically simple procedures. An alternative to this design is given. Reports on trials should contain information about pretrial routine and level of training of participants.