The Markov condition for a Bayesian network states that the any node in a Bayesian network is conditionally independent of its nondescendents, given its parents.

A node is conditionally independent of the entire network, given its Markov blanket.