This study examined the relative effectiveness of Bayesian model comparison methods in selecting an appropriate graded response (GR) model for performance assessment applications. Three popular methods were considered: deviance information criterion (DIC), conditional predictive ordinate (CPO), and posterior predictive model checking (PPMC). Using these methods, several alternative GR models were compared with Samejima's unidimensional GR model, including the one-parameter GR model, the rating scale model, simple- and complex-structure two-dimensional GR models, and the GR model for testlets. Results from a simulation study indicated that these methods appeared to be equally accurate in selecting the preferred model. However, CPO and PPMC can be used to compare models at the item level, and PPMC can also be used to compare both the relative and absolute fit of different models. (Contains 5 tables and 1 figure.)