We review some of the commonly used approximation methods to predict large-scale structure formation in modified gravity (MG) models for the cosmic acceleration. These methods are developed to speed up the often slow [Formula: see text]-body simulations in these models, or directly make approximate predictions of relevant physical quantities. In both cases, they are orders of magnitude more efficient than full simulations, making it possible to explore and delineate the large cosmological parameter space. On the other hand, there is a wide variation of their accuracies and ranges of validity, and these are usually not known a priori and must be validated against simulations. Therefore, a combination of full simulations and approximation methods will offer both efficiency and reliability. The approximation methods are also important from a theoretical point of view, since they can often offer useful insight into the nonlinear physics in MG models and inspire new algorithms for simulations.