Soil stabilization is widely embraced as a method to enhance unfavorable soil geotechnical properties, rendering them conducive for subgrade use in pavement construction. Enhancement of weak soil subgrade stands as a pivotal and paramount procedure in highway construction projects. Subgrade soil strength is measured based on its California bearing ratio (CBR) value. Conduction of CBR test for subgrade materials along the entire road length is tedious task. To address this challenge, the current study endeavors to forecast CBR value of black cotton soil stabilized with coal ash using artificial neural network (ANN) modeling. Experimental data regarding the stabilization of black cotton soils with varying proportions of industrial waste by-product, i.e., coal ash. Percentage of ash, maximum dry density, liquid limit, optimum moisture content, and plastic limit are utilized as input parameters, with the soaked CBR value being the target variable for the computational model. Network used for developing ANN model is feed-forward back propagation. In this study, the chosen ANN model adopts the Levenberg–Marquardt training algorithm and a 5-10-1 architecture, deemed the most suitable for analysis. The model employed to foresee CBR values of black cotton soil stabilized with coal ash demonstrates an impressive overall R value 0.9700 and minimal mean squared error (MSE) of 0.0235. In addition to the training data, 21 additional samples were employed to predict the CBR value using the generated model. Remarkably, the predictions closely aligned with laboratory results. The findings demonstrated that ANN prediction models are a reliable and effective approach for accurately forecasting the CBR value of stabilized black cotton soil using index properties as input variables.