Understanding the volume fractions of microstructure constituents such as ferrite, pearlite, bainite, and martensite in low-alloy steels is critical for tailoring mechanical properties to specific engineering applications. To address the complexity of these relationships, this study explores the use of artificial neural networks (ANNs) as a robust tool for predicting these microstructure constituents based on alloy composition, specific Jominy distance, and heat treatment parameters. Unlike previous ANN-based predictions that rely on the hardness after quenching as an input parameter, this study excludes hardness. The developed model relies on readily available input parameters, enabling accurate estimation of microstructure composition prior to heat treatment, which significantly improves its practicality for process planning, optimization, and reducing trial-and-error on industrial applications. Three different input configurations were tested to evaluate the predictive capabilities of ANNs, with results showing that the use of specific Jominy distance as an input variable enhances model performance. Furthermore, the findings suggest that specific Jominy distance could serve as a practical alternative to detailed chemical composition data in industrial applications. The predictions for ferrite, pearlite, and martensite were more accurate than those for bainite, which can be attributed to the complex nature of bainite formation.