Superparameterized climate models have been shown to improve both ends of the precipitation distribution. Yet even as cloud-resolving scales on the order of a few kilometers are reached, other parameterizations remain, including that of microphysical processes. In this study, we analyze precipitation extremes within the continental United States using a superparameterized model (SP-CAM) with four microphysics configurations. We vary the number of predicted moments, the character of the rimed ice species, i.e., graupel vs. hail, and the representation of raindrop self-collection and breakup. While the spatial structure of mean precipitation and of winter extremes is similar amongst the four schemes, greater spatial variation arises in the summer extremes. The single-moment scheme (Khairoutdinov and Randall, 2003) produces less precipitation in the mean but extremes comparable, if not higher, in magnitude to those of the other three, double-moment (Morrison et al., 2005), configurations. Using graupel as the rimed ice species results in more intense rainfall than does using hail. Increasing the size threshold at which raindrops’ breakup accelerates to counteract self-collection generally results in weaker extremes.