Causal inference—that is, determining the “root cause(s)” of an observed anomaly—is one of the most fundamental audit tasks. This study develops an analytical framework to formally model conditions present in many audit settings, and provides illustrations related to performing substantive analytical procedures. We examine four conditions not fully considered in prior research: multiple hypotheses about what may cause an anomaly, multiple items of evidence with varying diagnosticity, observed effects that may not be certain, and hypotheses sets that may not be exhaustive. The results reveal when the following phenomena should occur: (1) discounting or inflating of posterior probabilities, (2) superadditive probabilities of various causes, and (3) unchanged probability of a potential cause given evidence in support of a different cause. The analytical findings have implications for the design and interpretation of experimental auditing research, for educating novice auditors, and for potentially improving audit practice.