<p>Reliable prediction of tensile properties for additively manufactured materials is in urgent need for wider adoption of additive manufacturing in industry. This paper presents a unique and comprehensive study on utilizing machine learning techniques for predicting the ultimate tensile strength (UTS) of as-built Inconel 718 from selective laser melting (SLM), a predominant additive manufacturing (AM) technique for metallic materials. Seven representative machine learning (ML) models were selected to evaluate their predictive capabilities. A comprehensive dataset, which is comprised of SLM process parameters and UTS of as-built Inconel 718, was compiled by mining the literature. The hyperparameters tuning for the models was performed through extensive grid search, and leave-one-out cross-validation (LOOCV) was employed to reduce bias and ensure reliable performance of the models. It was found that the inclusion of volumetric energy density as an additional input variable to the regular SLM process parameters led to the best prediction performance. In this scenario, the best <i>R</i>² values in validating the seven ML models, namely, Linear Regression, XGBoost Regressor, Gradient Boosting Regressor, AdaBoost Regressor, Random Forest Regressor, Gaussian Process Regressor and Multi-layer Perceptron Regressor, were obtained as 5.62%, 72.19%, 71.56%, 73.03%, 69.52%, 70.11%, 67.23%, respectively. The most accurate prediction in testing against the new in-house SLM experiments was from XGBoost Regressor with 95.5% accuracy though the limited size of the test dataset may introduce some uncertainty. In the end, XGBoost and AdaBoost Regressor were recommended for predicting the UTS of Inconel 718 for high generalization in the cross-validation and high accuracy in the testing data.</p>