With the recent advancement in the field of machine learning, health synthetic data has become a promising technique to address difficulties with time consumption when accessing and using electronic medical records for research and innovations. However, health synthetic data utility and governance have not been extensively studied. A scoping review was conducted to understand the status of evaluations and governance of health synthetic data following the PRISMA guidelines. The results showed that if synthetic health data are generated via proper methods, the risk of privacy leaks has been low and data quality is comparative to real data. However, the generation of health synthetic data has been generated on a case-by-case basis instead of being scaled up. Furthermore, regulations, ethics, and data sharing of health synthetic data have primarily been inexplicit, although common principles for sharing such data do exist.