Brain-Computer Interface (BCI) is a rapidly developing technology that allows direct communications between the human brain and external devices, such as robotic arms and computers. Bayesian Networks is a powerful tool in machine learning for tackling with problems that requires understanding and modelling the uncertainty and complexity within complex system built by sub-modular components. Therefore, deploying Bayesian Networks in the application of Brain-Computer Interfaces becomes an increasingly popular approach in BCI research. This survey covers related existing works in relatively high-level perspectives, classifies the models and algorithms involved, and also summarizes the application of Bayesian Networks or its variants in the context of Brain-Computer Interfaces.