In this study, a unique multi-camera three-dimensional digital image correlation (3D-DIC) system was designed and applied to an engine dynamometer cell to measure the displacement and strain of the exhaust manifold while an engine was running in a durability test. In the engine dynamometer cell, the ambient temperature varies from 25°C to 80°C~100°C cyclically and the exhaust manifold experiences high temperatures up to 900°C with high frequency vibrations. In order to obtain reliable data under such conditions, two 3D-DIC systems were designed and set up in the engine dynamometer. One is a high-speed 3D-DIC system, consisting of cameras with a sampling rate of 1250 frames per second. It was used to measure the local displacement of the bolted joint in the exhaust manifold. The high-speed measurement system is able to record the behavior of the bolt during the thermal cycles. The other system is a high-resolution 3D-DIC system, consisting of two cameras with a resolution of five-mega-pixels. It was used to acquire the displacement and strain fields of the entire exhaust manifold. A Linear Variable Differential Transformer (LVDT) was used to assess the DIC data. A glass-ceramic cube with low thermal expansion was used to monitor the error of the 3D-DIC system caused by the change of the ambient temperature. With these two unique 3D-DIC systems, the deformation and strain of the exhaust manifold and the bolt were measured while the engine was running. In addition to the experimental test, the engine performance and the temperature change of the exhaust manifold were simulated by virtual analysis based on the engine test schedule and the dynamometer setup. The strain field and the displacement contour were calculated using a non-linear transient finite element (FE) method with a Global-Submodeling technique. The simulation results were compared to the experimental data. The FE model was validated based on the experimental data. This paper describes the unique 3D-DIC system setup in the engine dynamometer, theoretical explanation, experimental procedures, test data post-processing, virtual simulation method, correlation study, and the FE model improvement.