Water quality plays a key role in ensuring public health, maintaining sustainable agricultural development and the efficiency of industrial processes. This paper presents an approach to classifying water by categories of its use using a decision tree algorithm. The study is based on the analysis of physico-chemical and microbiological characteristics, including the concentration of calcium, magnesium, organic substances, phosphates, nitrites, sulfates, the level of alkalinity, residual chlorine and water oxidizability. The dataset used includes 1300 records evenly distributed among four categories: drinking, industrial, agricultural and industrial water. The constructed model achieved classification accuracy of 99.2%, which is confirmed by quality metrics and confusion matrix analysis. Visualization of the importance of the signs showed that the concentration of calcium, total organic matter and alkalinity are the most significant for the separation of water categories. The analysis of the main components confirmed the high separability of categories in the feature space. The results of the study demonstrate the effectiveness of machine learning for water classification and provide new opportunities for monitoring its quality. The model can be integrated into environmental analysis and water management systems, contributing to automation and improving the accuracy of monitoring the condition of water bodies.