Granular activated carbon (GAC) adsorption is frequently considered to control recalcitrant organic micropollutants (MPs) in both drinking water and wastewater. To predict full-scale GAC adsorber performance, bench- and/or pilot- scale studies are widely used. These studies have generated a wealth of MP breakthrough curves. The overarching aim of this research was to develop machine learning (ML) models from these data to predict MP breakthrough from adsorbent, adsorbate, and background water matrix properties. These models provide a simple and fast tool to predict GAC performance. To develop information for model calibration, MP breakthrough curves were collected from the peer-reviewed literature, research reports, and engineering reports. These data sets, which included results from rapid small-scale column tests (RSSCTs) and pilot/full-scale adsorbers, were analyzed to determine the bed volumes of water that could be treated until MP breakthrough reached ten percent of the influent MP concentration (BV10). The data set encompassed 43 MPs (including neutral and ionizable organic compounds), 3 GAC types by base material (18 unique GAC products), and 38 water matrices, including groundwater, surface water, and treated wastewater. Approximately 400 data sets were split into training, validation, and test sets. Seventeen candidate features, such as MP properties (Abraham parameters), background water matrix characteristics, and GAC properties, were explored in ML models to predict log-10-transformed BV10 (logBV10). BV10 values obtained from the resulting predictive model were highly correlated with experimentally determined BV10 values (coefficient of determination ~0.89 for logBV10 prediction), and the most effective model predicted BV10 with an absolute mean error of ~ 0.11 log units. Key drivers influencing BV10 prediction included the MP’s partitioning coefficient between air and hexadecane (Abraham parameter L); dissolved organic matter concentration in background water matrix; and the adsorbent’s point of zero charge (pzc). The model can be used to estimate GAC bed life and select effective GACs for the removal of MPs such as per- and polyfluoroalkyl substances (PFASs), pesticides, pharmaceuticals, and volatile organic compounds (VOCs) in a wide range of water types.