Advances in data analytics techniques allow auditors to process the entire population of transaction data to identify outliers (i.e., unusual/suspicious transactions) that are more likely to be subject to misstatement. However, these techniques often generate a large number of outliers, making it impractical for auditors to investigate them in their entirety when performing substantive tests. This study proposes a Multidimensional Audit Data Selection (MADS) framework that provides a systematic approach for auditors to use data analytics in the audit data selection process. The framework also addresses a common obstacle of applying data analytics to the entire population of data—dealing with a potentially large number of outliers. By identifying problematic items from the entire population using data analytics and then applying prioritization methodologies to the resulting items, this framework allows auditors to focus on items with a higher risk of material misstatement and ultimately enhance the effectiveness of the audit.