Summary. Despite the significant impact of fraud, forensic audit investigations are carried out in a fragmented and dispersed manner. Due to the nature of the topic, its study is confusing. In this regard, the aim is to characterise the empirical approach in forensic auditing that investigators have used to study fraud. The methodology was the descriptive-exploratory literature review, using the SCOPUS database. Search descriptors are identified, organise and analyse the results in the light of their methodological characteristics; in a descriptive review process, work is based on critical reading for review, organisation, labelling, integration and prioritisation; using perches, analysis and synthesis tables. The results show that the fraud study is addressed by individuals, documents, report imperfections, ICT, education, procedural innovation, experiments and simulations, interpretative language biases and the harms of crime. It is concluded that there are investigations aimed at identifying the reasons why individuals decide not to commit fraud and the compensation was identified as an alternative theoretical approach.