As a result of increasing highway construction and expansion, a corresponding need to increase traffic capacity in heavily populated areas, and ever-increasing constraints on available land for transportation use, there has been an increasing demand for alignment geometries and bridge configurations that result in more efficient use of available space. As a result of this demand, there has been a steady increase in the use of curved girder bridges over the past 30 years. Despites extensive research relating to the behavior of these types of structures, a thorough understanding of curved girder bridge response, especially relating to dynamic behavior, is still incomplete. To develop an improved, rational set of design guidelines, the Federal Highway Administration (FHWA) initiated the Curved Steel Bridge Research Project in 1992. As part of this project, FHWA constructed a full-scale model of a curved steel girder bridge at its Turner-Fairbank Structures Laboratory. This full-scale model made it possible to conduct numerous tests and collect a significant amount of data relating to the static behavior of a curved girder bridge. However, relatively little information has been available on the dynamic response of curved girder bridges and this type of information is needed before a complete design specification can be developed. The objective of this study was to develop a finite element model using SAP2000 that could be used for predicting and evaluating the dynamic response of a curved girder bridge. Models of the FHWA curved girder bridge were developed using both beam and shell elements and response information compared with experimental data and with analytical data from other finite element codes. The experimental data were obtained during dynamic testing of the full-scale bridge in the Turner-Fairbank Structures Laboratory and analytical response information was provided from finite element models of the bridge using ANSYS and ABAQUS. The primary focus of the study was the prediction of frequencies and mode shapes of the full-scale curved girder both with and without a deck. Both experimental and analytical frequencies and mode shapes were calculated and compared. Although the more refined ANSYS and ABAQUS models provided response data that compared more favorably with the experimental data, the SAP2000 models were found to be more than adequate for predicting the lower modes and frequencies of the bridge.