The paper introduces three-way decision rules based on the classical and decision-theoretic rough set models. In the classical rough set model, a concept is approximated by three regions: the positive region (objects definitely belonging to the concept), the boundary region (objects possibly belonging), and the negative region (objects definitely not belonging). These regions are used to construct three types of rules: positive (acceptance), boundary (abstention), and negative (rejection). The three-way decision approach allows for more nuanced decision-making by considering uncertainty and tolerance for errors. In the decision-theoretic rough set model, probabilistic measures are incorporated to account for the costs and risks associated with different decisions. The three-way decision rules are derived based on a loss function that quantifies the costs of different actions (acceptance, rejection, abstention). The decision-theoretic rough set model provides a systematic way to determine thresholds for these decisions based on the loss function and the probabilities of objects belonging to the concept. The three-way decision approach is closely related to Bayesian decision theory and hypothesis testing in statistics. It allows for a more practical application of rough set theory by incorporating uncertainty and cost considerations. The three-way decision rules are particularly useful in scenarios where decisions must be made under uncertainty, such as in medical diagnosis, information filtering, and statistical hypothesis testing. The paper also discusses the application of three-way decision rules in two-category classification. It introduces probabilistic rules for classification based on the three-way decision framework. These rules allow for more accurate and nuanced classification by considering the probabilities of objects belonging to each class and the associated costs of different decisions. The three-way decision approach provides a flexible framework for classification that can be adapted to different scenarios and applications.The paper introduces three-way decision rules based on the classical and decision-theoretic rough set models. In the classical rough set model, a concept is approximated by three regions: the positive region (objects definitely belonging to the concept), the boundary region (objects possibly belonging), and the negative region (objects definitely not belonging). These regions are used to construct three types of rules: positive (acceptance), boundary (abstention), and negative (rejection). The three-way decision approach allows for more nuanced decision-making by considering uncertainty and tolerance for errors. In the decision-theoretic rough set model, probabilistic measures are incorporated to account for the costs and risks associated with different decisions. The three-way decision rules are derived based on a loss function that quantifies the costs of different actions (acceptance, rejection, abstention). The decision-theoretic rough set model provides a systematic way to determine thresholds for these decisions based on the loss function and the probabilities of objects belonging to the concept. The three-way decision approach is closely related to Bayesian decision theory and hypothesis testing in statistics. It allows for a more practical application of rough set theory by incorporating uncertainty and cost considerations. The three-way decision rules are particularly useful in scenarios where decisions must be made under uncertainty, such as in medical diagnosis, information filtering, and statistical hypothesis testing. The paper also discusses the application of three-way decision rules in two-category classification. It introduces probabilistic rules for classification based on the three-way decision framework. These rules allow for more accurate and nuanced classification by considering the probabilities of objects belonging to each class and the associated costs of different decisions. The three-way decision approach provides a flexible framework for classification that can be adapted to different scenarios and applications.