This paper addresses the challenge of extracting classification rules to identify patterns within relational databases. It introduces an information and statistical method designed to create a set of statistically significant classification rules. The proposed approach utilizes the Kullback–Leibler information measure and chi-squared statistic to assess the statistical significance of the rules. By focusing on quantitative characteristics such as support, accuracy, and completeness, the method ensures that the extracted rules are both meaningful and reliable. The algorithm developed for this purpose systematically filters out insignificant rules in the preliminary stages, enhancing the efficiency and accuracy of the rule extraction process. This research contributes to the field of data mining by providing a robust framework for discovering dependencies between database attributes, ultimately facilitating more informed decision-making in various applications.