One characteristic of a Geographic Information System (GIS) is that it addresses the necessity to handle a large amount of data at multiple scales. Lands span over an area greater than 15 million km2 all over the globe and information types are highly variable. In addition, multi-scale analyses involve both spatial and temporal integration of datasets deriving from different sources. The currently worldwide used system of latitude and longitude coordinates could avoid limitations in data use due to biases and approximations. In this article a fast and reliable algorithm implemented in Arc Macro Language (AML) is presented to provide an automatic computation of the surface area of the cells in a regularly spaced longitude-latitude (geographic) grid at different resolutions. The approach is based on the well-known approximation of the spheroidal Earth's surface to the authalic (i.e. equal-area) sphere. After verifying the algorithm's strength by comparison with a numerical solution for the reference spheroidal model, specific case studies are introduced to evaluate the differences when switching from geographic to projected coordinate systems. This is done at different resolutions and using different formulations to calculate cell areas. Even if the percentage differences are low, they become relevant when reported in absolute terms (hectares).