In high-throughput screening, an extremely large chemical space is searched to discover materials with desired properties. For electrolytes with potential applications in lithium batteries, electrochemical stability at the operating voltage of electrodes or the ability to form a stable passivation layer on electrode surfaces are critical. Quantum chemistry (QC) models are capable of accurate prediction of electrochemical properties relevant to electrolyte design given a sufficient level of theory, e.g. correlated methods. However, high-fidelity QC models also have a high computational cost. Therefore, to expedite high-throughput screening of potential electrolytes, lower fidelity methods such as density functional theory or semi-empirical methods are often used for property prediction. In this talk, we will discuss recent work on the use of multi-fidelity surrogate models in high-throughput screening of battery electrolytes. Multi-fidelity surrogate models provide a mathematical framework to combine the predictions of material properties obtained from multiple models. In multi-fidelity high-throughput screening, a lower-fidelity model is used to rapidly explore the chemical space in combination with far fewer evaluations of an expensive high-fidelity model. The multi-fidelity surrogate model serves to discover the correlation structure between the two models from available data and to correct low-fidelity model predictions. Subsequently, the multi-fidelity surrogate model is used to guide the search for electrolytes with desired properties.