To set the framework for Bayesian SEM, we believe it is important to start first with description of the Bayesian approach. The literature currently lacks such description, not only within the context of SEM but within other modelling approaches. We focus on the basic ideas of Bayesian inference for both model estimation and model comparison.