Rational theories of diagnostic reasoning assume that the reasoner’s goal is to infer the conditional probability of a cause given an effect from the available data. Typically, diagnostic reasoning is modeled within a statistical inference framework, with Bayes’ rule applied to the obtained covariation information serving as the normative standard. This chapter analyzes diagnostic reasoning from the perspective of causal induction, using the framework of causal Bayes net theory to instantiate different accounts of rational diagnostic reasoning. These approaches elucidate the relevant kinds of inputs, computations, and outputs by differentiating between parametric causal models and observable contingency information. A particularly interesting feature of these accounts is that they can include predictions that systematically deviate from the traditional, purely statistical norm. The analyses highlight key issues for constructing a rational theory of diagnostic reasoning and the experimental study of human rationality.