The land surface model (LSM) ISBA-A-gs (Interactions between Soil, Biosphere and Atmosphere, CO2-reactive) is specifically designed to simulate leaf stomatal conductance and leaf area index (LAI) in response to climate, soil properties, and atmospheric carbon dioxide concentration. The model is run at the global scale, forced by the GSWP-2 meteorological data at a resolution of 1¿ for the period of 1986--1995. We test the model by comparing the simulated LAI values against three satellite-derived data sets (ISLSCP Initiative II data, MODIS data and ECOCLIMAP data) and find that the model reproduces the major patterns of spatial and temporal variability in global vegetation. As a result, the mean of the maximum annual LAI estimates of the model falls within the range of the various satellite data sets. Despite no explicit representation of phenology, the model captures the seasonal cycle in LAI well and shows realistic variations in start of the growing season as a function of latitude. The interannual variability is also well reported for numerous regions of the world, particularly where precipitation controls photosynthesis. The comparison also reveals that some processes need to be improved or introduced in the model, in particular the snow dynamics and the treatment of vegetation in cultivated areas, respectively. The overall comparisons demonstrate the potential of ISBA-A-gs model to simulate LAI in a realistic fashion at the global scale.