Microphysical processes in cold clouds which act as sources or sinks of hydrometeors below 0 .sup." C control the ice crystal number concentrations (ICNCs) and in turn the cloud radiative effects. Estimating the relative importance of the cold cloud microphysical process rates is of fundamental importance to underpin the development of cloud parameterizations for weather, atmospheric chemistry, and climate models and to compare the output with observations at different temporal resolutions. This study quantifies and investigates the ICNC rates of cold cloud microphysical processes by means of the chemistry-climate model EMAC (ECHAM/MESSy Atmospheric Chemistry) and defines the hierarchy of sources and sinks of ice crystals. Both microphysical process rates, such as ice nucleation, aggregation, and secondary ice production, and unphysical correction terms are presented. Model ICNCs are also compared against a satellite climatology. We found that model ICNCs are in overall agreement with satellite observations in terms of spatial distribution, although the values are overestimated, especially around high mountains. The analysis of ice crystal rates is carried out both at global and at regional scales. We found that globally the freezing of cloud droplets and convective detrainment over tropical land masses are the dominant sources of ice crystals, while aggregation and accretion act as the largest sinks. In general, all processes are characterized by highly skewed distributions. Moreover, the influence of (a) different ice nucleation parameterizations and (b) a future global warming scenario on the rates has been analysed in two sensitivity studies. In the first, we found that the application of different parameterizations for ice nucleation changes the hierarchy of ice crystal sources only slightly. In the second, all microphysical processes follow an upward shift in altitude and an increase by up to 10 % in the upper troposphere towards the end of the 21st century.