The evaluation of public engagement health festivals is of growing importance, but there has been no synthesis of its practice to date. We conducted a systematic review of evidence from the evaluation of health-related public engagement festivals published since 2000 to inform future evaluation. Primary study quality was assessed using the Mixed Methods Appraisal Tool. Extracted data were integrated using narrative synthesis, with evaluation methods compared with the Queen Mary University of London public engagement evaluation toolkit. 407 database records were screened; eight studies of varied methodological quality met the inclusion criteria. Evaluations frequently used questionnaires to collect mixed-methods data. Higher quality studies had specific evaluation aims, used a wider variety of evaluation methods and had independent evaluation teams. Evaluation sample profiles were often gender-biased and not ethnically representative. Patient involvement in event delivery supported learning and engagement. These findings and recommendations can help improve future evaluations. (Research Registry ID reviewregistry1021).