Partial Least Squares (PLS) is a statistical technique that is widely used in the Information Systems discipline to estimate statistical models with structural equations and latent variables. While PLS does not provide a statistical test of model fit to data, its proponents have suggested a set of criteria that good PLS models should fulfill. Conversely, when a model does not satisfy these criteria, it would be judged a bad model. In this paper, we report on the results of a simulation study to examine to what extent the proposed model quality criteria are able to identify misspecified models.