It is necessary to develop a flexible and extensible watershed modeling framework with the support of parallel computing to conduct long-term high-resolution simulations over large areas with diverse watershed characteristics. This paper introduced an open-source, modular, and parallelized watershed modeling framework called SEIMS (short for Spatially Explicit Integrated Modeling System) to meet this need. First, a flexible modular structure with standard interfaces was designed, in which each module corresponds to one simulation algorithm for a watershed subprocess. Then, a parallel-computing middleware based on an improved two-level parallel computing approach was constructed to speed up the computational efficiency. With SEIMS, users can add their own algorithms in a nearly serial programming manner and construct parallelized watershed models. SEIMS also supports model level parallel computation for applications which need numerous model runs. The effectiveness and efficiency of SEIMS were illustrated through the simulation of streamflow in the Youwuzhen watershed, Southeastern China.