The paper focuses on the empirical evidence on the use and usefulness of performance measures in the public sector. It begins with consideration of the features of the public sector which make the use of performance measures complex: the issues of multiple principals and multiple tasks. It discusses the form that performance measures may take, the use made of these measures, and the responses that individuals may make to them. Empirical examples from the fields of education and health, with a focus on the USA and UK, are examined. There is clear evidence of responses to such measures. Some of these responses improve efficiency, but others do not and fall into the category of 'gaming'. Generally, there has been little assessment of whether performance measures bring about improvements in service. The paper ends with consideration of how such measures should be used and what measures are useful to collect.