A 3-d numerical model is being used for the development of improved parametrizations of convective transport processes in climate and numerical weather-prediction models. In common with similar cloud-resolving models, it uses a 3-phase bulk-water parametrization of the microphysical processes. The comparison of results from such models with observed case-studies represents an important step in their development and a key activity of the Global Energy and Water Cycle Experiment Cloud System Study.This paper presents observations and model results from the case of a weakly convective precipitating cloud that occurred on 22 June 1990 over central southern England. The cloud was observed by both an instrumented research aircraft and a dual-polarization radar. The combination of these measurements provides information on the dominant precipitation particle types, their density, number concentration, and characteristic size, and enables the identification of key microphysical processes. The observations were used to provide guidance on the numerical values of some of the variable parameters in the microphysics scheme. Observed values of No,G, the intercept parameter for graupel, lay in the range lo4 to lo7 m-4. The comparison of observed radar reflectivities with those calculated from in situ particle size spectra suggests that the density of rimed precipitation particles decreased from around 0.5 g cm-3 in the most intense precipitation to around 0.1 g A number of model sensitivity studies were performed to examine the impact of changes in these parameters on the model cloud and precipitation fields. For this case-study, it was found that the total surface precipitation was generally insensitive to such changes, varying by less than 10%. Reductions of 35% occurred for model tuns in which the graupel fall speed was reduced and for a model run with only 'warm-rain' microphysics. Much larger variations in the range 50-100% occurred in the total cloud ice and snow contents.The modelling study identifies a number of differences between runs which could not be resolved owing to the lack of in situ data in certain regions of the cloud. Some of the requirements for improved observational methods and procedures are discussed.towards the cloud edges.