Dynamic Bayesian networks (DBNs) as an extension of traditional Bayesian networks have recently been paid great concern to environmental modeling to capture dynamic processes and support feedback loops. However, the applications of DBNs in environmental modeling are still scarce and challenging. There are no reviews found in the literature to explore the potential and application of DBNs in the environmental science fields so far. This review is to illustrate how DBNs are applied in the environmental modeling and management. The overview of DBNs is performed from January 1990 to December 2021 in the Web of Science search related to Environmental Sciences. Only 5.69% of the total publications have been used in this item. The application fields, model types, model aims, model learning, model spatial and temporal scale, model validation, and software are discussed. The pros and cons analysis highlights the advantages and disadvantages of DBNs in the environmental modeling. The current DBNs research focuses mainly on environment applications through expert knowledge for learning and validation of models due to less available support of the popular commercial software packages and data availability limitation. The powerful potential of DBNs is yet unexploited. Some challenges including timeline discretization, modeling accuracy in medium- and long-term scale, modeling of spatial dependencies and interactions, and development of algorithms and software have to be considered in future research.