Broadly neutralizing antibodies (bNAbs) targeting the HIV-1 envelope glycoprotein (Env) have promising utility in prevention and treatment of HIV-1 infection, and several are currently undergoing clinical trials. Due to the high sequence diversity and mutation rate of HIV-1, viral isolates are often resistant to specific bNAbs. Currently, resistant isolates are commonly identified by time-consuming and expensive in vitro neutralization assays. Here, we report machine learning classifiers that accurately predict resistance of HIV-1 isolates to 33 bNAbs. Notably, our classifiers achieved an overall prediction accuracy of 96% for 212 clinical isolates from patients enrolled in four different clinical trials. Moreover, use of gradient boosting machine - a tree-based machine learning method - enabled us to identify critical features, which had high accordance with epitope residues that distinguished between antibody resistance and sensitivity. The availability of an in silico antibody resistance predictor should facilitate informed decisions of antibody usage and sequence-based monitoring of viral escape in clinical settings.