Evaluations are key to learning and accountability. We assessed the methodological quality of 37 randomly selected programme evaluations from 5 major global health funders. Two researchers rated each evaluation for relevance, validity, and reliability and met to resolve discrepancies. Most evaluations asked questions relevant to the health programme, but less than 40 per cent of impact evaluations and less than 10 per cent of performance evaluations used relevant data, followed accepted social science methods for sampling, or had high analytical validity and reliability. There is a need to improve the methodological quality of programme evaluations.