For the past decade, considerable research effort has been devoted toward computationally identifying and experimentally verifying single phase, high-entropy systems.  However, predicting the resultant crystal structure(s) “in silico” remains a major challenge. Previous studies have primarily used density functional theory to obtain correlated parameters and fit them to existing data, but this is impractical given the extensive regions of unexplored composition space and considerable computational cost. A rapidly developing area of materials science is the application of machine learning to accelerate materials discovery and reduce computational and experimental costs. Machine learning has inherent advantages over traditional modeling, owing to its flexibility as new data becomes available and its rapid ability to construct relationships between input data and target outputs. In this article, we propose a novel high-throughput approach, called “ML-HEA”, for coupling thermodynamic and chemical features with a random forest machine learning model for predicting the solid solution forming ability. The model can be a primary tool or integrated into existing alloy discovery workflows. The ML-HEA method is validated by comparing the results with reliable experimental data for binary, ternary, quaternary, and quinary systems. Comparison to other modeling approaches, including CALPHAD and the LTVC model, are also made to assess the performance of the machine learning model on labeled and unlabeled data. The uncertainty of the model in predicting the resultant phase of each composition is explored via the output of individual predictor trees. Importantly, the developed model can be immediately applied to explore material space in an unconstrained manner, and is readily updated to reflect the results of new experiments.