Expansive soils, known for their significant volume changes due to moisture variations, pose considerable challenges in geotechnical engineering. These clayey soils, including vertisols and montmorillonite clays, swell upon wetting and shrink upon drying, leading to potential damage to structures such as pavements and single-story buildings. Originating from the weathering of both igneous and sedimentary rocks, expansive soils can cause severe structural issues due to their volumetric deformations. Traditional methods for assessing swelling potential involve labour-intensive and costly laboratory tests, including Atterberg limits, free swell tests, and X-ray diffraction. However, advancements in artificial neural networks (ANNs) present a promising alternative. ANNs, inspired by the human brain’s structure and learning processes, use interconnected nodes to model complex relationships between soil properties and swelling behaviour. This paper examines the use of ANN to predict swell pressure based on input parameters such as liquid limit, plasticity index, dry density, and free swell. By leveraging ANN models, the study aims to offer a more efficient and cost-effective approach for predicting swelling pressures compared to traditional methods. The effectiveness of the ANN models is evaluated by comparing predicted results with experimental data, demonstrating their potential to enhance the prediction and management of expansive soil behaviour. This approach can significantly improve decision-making in the design and construction of foundations and structures on expansive soils.