Abstract: We present the GBRAMS project that applies grid computing to speed-up the generation of a regional model (BRAMS) climatology for Brazil. The project stresses the grid concept as the driver of slaves PC clusters. We used three grid proposals (Globus, OurGrid and CIGRI), reaching similar execution times and cluster occupancy. By dividing Brazil in three independent geographic regions and by using a three member ensemble we were able to keep all clusters busy and reduce the total execution time of obtaining climatology by the number of member clusters.