<p>Soil liquefaction, defined by the abrupt reduction in strength and rigidity of saturated soils when subjected to cyclic or static loading, is a critical issue for the risk assessment community. This phenomenon has the potential to result in substantial soil failure and structural damage, particularly during seismic activities or periods of heavy rainfall. Traditional predictive methodologies have predominantly relied upon empirical correlations and field testing. However, the advent of machine learning (ML) has introduced a novel and promising approach for the prediction of liquefaction behavior, capable of capturing the inherent nonlinearity of this process. Existing ML methods are mostly founded on field experimental data, and there exists a gap in the development of models using laboratory-based experimental data. The objective of this research is to leverage the predictive prowess of ML methodologies to enhance soil liquefaction prediction. This is achieved by employing a comprehensive dataset, which has been meticulously curated from a variety of laboratory experiments. The tests include ring shear, triaxial, direct simple shear, and torsional shear tests. The dataset, comprising over 3000 laboratory test cases, was subjected to rigorous preprocessing and was subsequently partitioned into training and testing subsets. The core focus of this study is the formulation and rigorous evaluation of multiple ML algorithms, specifically Random Forest (RF), Extreme Gradient Boosting (XGBoost), Logistic Regression with Threshold Pushing (LRTP), and Support Vector Machines (SVM). These models are meticulously calibrated to predict the propensity for liquefaction. The efficacy of the models was evaluated using a suite of metrics, including accuracy, precision, recall, and F1-score. The XGBoost algorithm demonstrated superior performance, achieving an overall accuracy of 84%, with a test accuracy of 76% (95% confidence interval, CI at 73–80%), and precision, recall, and F1-score all at 76% (95% CI at 73–80%). Collectively, tree-based ML methods, specifically RF and XGBoost, emerged as the optimal choices for the prediction of soil liquefaction. The findings of this study highlight the proficiency of ML models, particularly XGBoost, in evaluating soil liquefaction potential with a high degree of accuracy, based on data derived from laboratory tests. This research posits that ML methodologies can offer a nuanced and precise evaluation of liquefaction risk, serving as an indispensable instrument in the mitigation of hazards linked to seismic or heavy rainfall events.</p>