This study examines the deep convection populations and mesoscale convective systems (MCSs) simulated in the DYAMOND (DYnamics of the atmospheric general circulation modeled on non-hydrostatic domains) winter project. A storm tracking algorithm is applied to six DYAMOND simulations and a global high-resolution satellite cloud and precipitation data set for comparison. The simulated frequencies of tropical deep convection and organized convective systems vary widely among models and regions, although robust MCSs are generally underestimated. The diurnal cycles of MCS initiation and mature stages are well simulated, but the amplitudes are exaggerated over land. Most models capture the observed MCS lifetime, cloud shield area, rainfall volume and movement speed. However, cloud-top height and convective rainfall intensity are consistently overestimated, and stratiform rainfall area and amount are consistently underestimated. Possible causes for the model differences compared to observations and implications for future model developments are discussed.