The MAGIC field campaign, conducted by the DOE Atmospheric Radiation Measurement (ARM) program in 2012 and 2013, provided comprehensive measurements of atmospheric properties over the northeast Pacific, emphasizing cloud properties in diverse regimes and the transitions between them. We use these data to evaluate simulations with a mesoscale model, WRF, in a quasi-operational configuration. The MYNN-EDMF boundary layer and shallow cloud scheme has been recently updated to improve its representation of subgrid clouds as well as vertical mixing in general. Here we describe the model performance in stratocumulus, transitional, and trade cumulus regimes. An ensemble of initializations from the ERA5 reanalysis is used to evaluate the significance of discrepancies between the model and observations. For MAGIC leg15A, the model locates the stratocumulus to cumulus transition well; however, the cloud base in stratocumulus is too low. Some of the error is inherited from ERA5. Vertical grid spacing of approximately 100 m in the cloud layer is necessary to produce good simulations, indicating the existence of a vertical gray zone. Interpretation of the transition process is complicated by the fact that the ship tracks are not Lagrangian. Importantly, decoupled and multi-layer soundings are common. In the cumulus regime, the size of cloud clusters is influenced by the horizontal grid spacing (baseline 13 km) and by the strength of the subgrid parameterization. This “gray zone” behavior calls into question conventional methods for scale awareness, which weaken parameterizations at smaller scales. In addition to the ship-based MAGIC data, we use satellite data to provide additional spatial context.