The provision of municipal services is contingent on appropriately managing municipal finances. Understanding the source and extent of expenditures that exceed the budgeted amount is essential for effective municipal financial management. The expenditures damaging to the municipality’s operation are irregular, unauthorized, and wasteful. This chapter uses unaudited financial statements from South African municipalities to detect unauthorized expenditures represented in the same entity’s audited financials. We provide a first-in-the-literature application of Bayesian machine learning to analyze the link between municipal financial performance measures (as reflected by financial ratios) and the prevalence of unauthorized expenditure. By utilizing a Bayesian framework with an automatic relevance determination prior, we also highlight the performance metrics that are the most relevant for detecting unauthorized expenditure using financial ratios from unaudited financial statements as inputs. This work provides tools that auditors could use for risk-rating municipal accounts for unauthorized expenditures, and it is easily extendable to other sets of unlawful expenditures, such as irregular and wasteful expenditures.