Fraudulent financial reporting is a matter of great social and economic concern. Managers may distort financial statements so as to present their companies more favorably to investors or creditors. On the other hand, auditors are the ones who are expected to identify fraudulent financial statements; however several limitations on auditors’ performance are reported. Therefore, decision support systems that can support auditor’s characterization and identify fraudulent statements can be of great benefit for the auditor himself but also for all stakeholders. In this work, we investigate the use of a swarm intelligence technique, mimicking the functions of natural ants for fraudulent financial statements detection. More specific, an Ant miner classifier is applied on a real dataset arising from the Athens Stock Exchange firms. The method is compared against commonly used statistical and machine learning classifiers. The obtained results are promising. Apart from the high classification accuracy, Ant miner offers interpretation ability, by extracting compact rules regarding the decisions made, which is a further advantage.