Abstract. The land-surface parameters required as input to a GCM grid box (typically a few degrees) are often set to be those of the dominant vegetation type within the grid box. This paper discusses the use and effect of aggregation rules for specifying effective values of these land cover parameters by taking into account the relative proportion of each land-cover type within each individual grid box. Global land-cover classification data at 1 km resolution were used to define Biosphere Atmosphere Transfer Scheme (BATS) specific aggregate (using aggregation rules) land-cover parameters. Comparison of the values of the aggregate parameters and those defined using the single dominant vegetation type (default parameters) shows significant differences in some regions, particularly in the semi-desert and in forested regions, e.g. the Sahara Desert and the tropical forest of South America. These two different sets of parameters were used as input data for two 10-year simulations of the NCAR CCM3 model coupled to the BATS land-surface scheme. Statistical analyses comparing the results of the two model runs showed that the resulting effects on the land-surface diagnostics are significant only in specific regions. For example, the sensible heat flux in the Sahara Desert calculated for the aggregate parameter run increased due to the marked increase in the minimum stomatal resistance and the decrease in fractional vegetation cover in the aggregate parameters over the default parameters. The modelled global precipitation and surface air temperature fields were compared to observations: there is a general improvement in the performance of the aggregate parameter run over the default parameter run in areas where the differences between the aggregate and default parameter run are significant. However, most of the difference between the modelled and observed fields is attributable to other model deficiencies. It can be concluded that the use of aggregation rules to derive land-surface parameters results in significant changes in modelled climate and in some improvements in the land-surface diagnostics in selected regions. There is also some evidence that there is a response in the global circulation pattern, which is a focus of further work.