Curved bridges are subject to combined loading conditions resulting from complex earthquake ground motions coupled with irregular geometry and asymmetry of the bridge structure. This paper describes a part of the CABER project that compares the experimental results of a curved bridge under multidirectional earthquake loading with its un-calibrated numerical predictions. The numerical predictions are obtained from an analytical model established using the Zeus-NL analysis platform. The experimental results are obtained from a hybrid experimental/analytical test performed at the NEES MUST-SIM Facility at the University of Illinois at Urbana-Champaign. In the hybrid test, the bridge is substructured into three analytical (one) and experimental (two) modules. Disagreements shown between the experimental and numerical results highly indicate the inadequacy of the existing analytical solutions. Numerical model assumptions and inaccuracy are identified and methods of numerical model calibration are proposed.