Three-dimensional vadose zone models are used more and more for solving hydrological problems on a broad range of scales with large amount of nodes. Currently, the problems we can solve in reasonable computational time may have up to 5 · 106 nodes. However, distributed models may need up to 1010 nodes to properly predict flow and transport at the watershed scale. The speed and efficiency of current flow and transport models therefore need to be improved. The parallelization of the code is one possible way to decrease the computational time by distributing a complex large geometry problem over multiple processors working in parallel. This is the solution we implemented by developing PARSWMS, a parallelized version of SWMS_3D (Simunek et al., 1995). The objective of this technical note is to describe the PARSWMS model, test its reliability, and show its performance and efficiency compared with single processor runs.