Heat capacity is an important and fundamental physicochemical property of ionic liquids (ILs). Here, a new class of quantum chemical descriptor, namely electrostatic potential surface area (SEP) descriptor, is employed to predict the heat capacity of ILs. In this study, 2416 experimental data points (254.0-1805.7 J mol-1 K-1) covering a wide temperature range (223.1-663 K) were employed. Multiple linear regression (MLR) and extreme learning machine (ELM) are applied to establish the linear and nonlinear models based on the SEP descriptors, respectively. The obtained six-parameter models show good predictive performance. The R2 of the linear MLR model is 0.988 for the entire set, while the ELM model has a higher value of R2=0.999, indicating the robustness of the nonlinear model. The results suggest that the SEP descriptors are closely related to the heat capacity of ILs and can be potentially used to predict the properties of ILs.