Efforts to transform health systems constitute social experiments on a population. Like clinical research, they deploy measures that are unproven in the context of the reform, and they often impose significant risks on some people in order to achieve a social goal: the improvement of health delivery. The rationale for proactively evaluating clinical experimentation on human subjects also applies to these social experiments. We used the "benchmarks of fairness" methodology to illustrate the elements such an evidence-based review should encompass, leaving open the question of who should perform it. The review must include the ethical objectives of reform, namely, an integrated approach to equity, accountability, and efficiency; the fit between measures taken and these objectives; and the governance of the reform.