The simulation of monsoons remains a challenging problem given the complexity of the multi-scale interactions and the modulating influences that operate on a broad range of time scales. Recent studies have demonstrated an improvement in the simulation of monsoons in CMIP5 relative to CMIP3, though many outstanding issues remain. For example, although mean state biases have been reduced in amplitude, the spatial error pattern remains virtually unchanged between the two vintages of experimentation. Systematic errors in the onset time of monsoons indicate that the simulated atmosphere-land-ocean interactions are not responding properly to the annual cycle of solar forcing. As such, the CMIP5 models have early monsoon onset over the Sahel and the North American domain, and late monsoon onset over India, the South American domain, and the Gulf of Guinea. This indicates that a regional process study approach is warranted for improving our understanding of atmosphere-land-ocean interactions inherent to monsoon development and onset. The simulation of intraseasonal variability remains a grand challenge problem, especially given its importance for initiating monsoon onset and association with precipitation extremes. Experimental prediction of intraseasonal variability suggests forecast skill to 3 weeks, with the potential for increased skill with the implementation of different projection basis functions for boreal summer versus boreal winter. Despite improvements in the simulation of the El Nino/Southern Oscillation, the interannual monsoon teleconnections are sensitive to biases in the regional rainfall that can compromise the response to the remote forcing. At best, interdecadal variations of Sahel rainfall are qualitatively captured in CMIP5, with the amplitude of the 1970s-1980s drought strongly underestimated. Conversely, mechanisms for observed interdecadal trends in rainfall over East Asia and northern Australia are yet to be understood, and these trends remain to be simulated. Higher horizontal resolution has been beneficial in the representation of orographic rainfall, as well as larger-scale aspects of the circulation (e.g., Baiu front), though improvements to model physics are most essential for improving the simulation and prediction of monsoons.