We develop a systematic coarse-grained (CG) model for methylcellulose polymers, including random copolymers with compositions representative of modeling commercial METHOCEL A polymer, using one CG bead per monomer. We parametrize our CG model using the RDFs from atomistic simulations of short methylcellulose oligomers, extrapolating the results to long chains. Using a LJ 9-6 potential, the CG model captures the effect of monomer substitution type and temperature observed in detailed atomistic simulations. We use dissociation free energy to validate our CG model against the atomistic model. We then use this CG model to simulate single chains up to 1000 monomers long, and we calculate persistence lengths for a selection of homogeneous and heterogeneous methylcellulose chains, which show good agreement with experimental results. Interestingly, simulations of 600-mer heterogeneous chains show a collapse transition at 50 degrees C and form a stable ring structure with outer diameter around 14 nm. This structure appears to be a precursor to fibril structure reported in a recent study of methylcellulose gels [Biomacromolecules 2013, 14, 2484].