The Massively Parallel Mesoscale Model (MPMM) is a distributed memory parallel implementation of the fifth-generation Penn State/NCAR Mesoscale Model (MM5) [1][2]. The parallel implementation of the MM5 is described in ``Parallel Implementation, Validation, and Performance of MM5'' [3]. The document you are reading describes how to acquire a copy of the code and parallel run-time system, configure its compile- and run-time settings, compile, and run the code. A section is also provided with information about building and running the model on the the Argonne National Laboratory IBM SP.