The use of probability for structural engineering applications has fallen into disfavour in recent years because of continuing debates about the modelling of human cognition, the assessment of marginal and conditional probabilities, the computational intractability of Bayesianism, and the so-called closed-world assumption. In this paper the validity of these arguments is examined and the use of the causal or Bayes network in estimating the reliability index for existing structures is proposed. The Bayes network was introduced in the artificial intelligence community as a method for propagating uncertainty through an inference network. Specifically, it is postulated that a structure can be modelled as a causal network