Proper Orthogonal Decomposition (POD) is applied to analyze the coherent structure of three-dimensional cylinder wake flow. The flow field data, such as velocity and pressure as functions of time, was obtained by the incompressible CFD analysis. The resulting CFD data was then used to determine eigenvalues, POD modes, and time coefficients through POD process. The flow field was approximately reconstructed using some of lower POD modes. The three-dimensional field reconstructed using the low-order model was found to be in good agreement with the original. This verifies that low-dimensional modeling of complex flow fields is fully possible.