Abstract: Current research on fatigue life prediction of steel wire ropes is constrained by experimental data and empirical formulas, posing significant limitations and uncertainties that impede its practical applications. To address this challenge, this study introduces a machine learning approach based on the extreme gradient boosting (XGBoost) algorithm. Initially, a finite element model of steel wire ropes was developed, incorporating basic geometric features as well as critical surface defect variables. This model enables precise simulation of stress distribution and deformation behavior under various tensile loads. Grounded on this foundation, load spectra were applied to estimate fatigue life, generating a comprehensive dataset that can reveal the profound impact of surface defects on fatigue life. Subsequently, the dataset was systematically analyzed using a Bayesian-optimized XGBoost algorithm, therefore developing a rapid and accurate prediction model for the tensile fatigue life of steel wire ropes. Compared to traditional machine learning algorithms such as linear regression and random forests, this model demonstrates superior prediction performance with a determination coefficient up to 0.988 1. This study not only provides a novel approach for predicting the fatigue life of steel wire ropes but also offers valuable support for their safe use and maintenance, highlighting its significant practical importance.