This paper explores what motivates improved health care performance. Previously, many have thought that performance would either improve via choice and competition or by relying on trust and altruism. But neither assumption is supported by available evidence. So instead we explore a third approach of reciprocal altruism with sanctions for unacceptably poor performance and rewards for high performance. These rewards and sanctions, however, are not monetary, but in the form of reputational effects through public reporting of benchmarking of performance. Drawing on natural experiments in Italy and the United Kingdom, we illustrate how public benchmarking can improve poor performance at the national level through 'naming and shaming' and enhance good performance at the sub-national level through 'competitive benchmarking' and peer learning. Ethnographic research in Zambia also showed how reputations count. Policy-makers could use these effects in different ways to improve public services.