Usually, data quality is defined by multiple attributes that allow classifying the output data (such as completeness, freshness, and accuracy) or the methods exploiting these data (such as dependability, performance, and protection). Among the suggested quality attributes, we will discuss one of the principal categories: data accuracy. Scientific experiments, decision–making, and data retrieval are examples of situations that require a formal evaluation approach to data accuracy. The evaluation approach should be adaptable to distinct understandings of data accuracy and distinct end-user expectations. This study investigates data accuracy and defines dimensions and metrics that affect its evaluation. The investigation of data accuracy generates problems in the user expectation specification and database quality models. This work describes our proposed approach for data accuracy evaluation by defining an evaluation algorithm that considers the distribution of inaccuracies in database relations. The approach decomposes the query output in accordance with data accuracy, labels every part with its accuracy value, and addresses the possibility of enforcing data accuracy by using these values. This study mainly contributes by proposing an explicit evaluation of quality attributes of data accuracy, a formal evaluation approach to data accuracy, and suggesting some improvement actions to reinforce data accuracy. © 2022 Slovene Society Informatika. All rights reserved.