RNA and RNA-protein complexes have recently been intensively studied in experiments, but the corresponding molecular simulation work is much less abundant, primarily due to its large system size and the long time scale involved. Here, to overcome these bottlenecks, we develop a coarse-grained (CG) structure-based simulation model for RNA and RNA-protein complexes and test it for several molecular systems. The CG model for RNA contains three particles per nucleotide, each for phosphate, sugar, and a base. Focusing on RNA molecules that fold to well-defined native structures, we employed a structure-based potential, which is similar to the Go-like potential successfully used in CG modeling of proteins. In addition, we tested three means to approximate electrostatic interactions. Many parameters involved in the CG potential were determined via a multiscale method: We matched the native fluctuation of the CG model with that by all-atom simulations for 16 RNA molecules and 10 RNA-protein complexes, from which we derived a generic set of CG parameters. We show that the derived parameters can reproduce native fluctuations well for four RNA and two RNA-protein complexes. For tRNA, the native fluctuation in solution includes large-amplitude motions that reach conformations nearly corresponding to the hybrid state P/E and EF-Tu-bound state A/T seen in the complexes with ribosome. Finally, large-amplitude modes of ribosome are briefly described.