Plant genotype/variety evaluation is an integral part of plant breeding research. In the advanced stages of this research, multilocation trials become important as a tool to assess variety yield and stability. Such trials do, however, produce a large amount of data that can be difficult to condense and interpret. A recently developed software application, GGEbiplot, allows for the graphical inspection of two-way data in an interactive manner. The objective of this research was to investigate the utility of GGEbiplot by analyzing lint yield data from the 1997-2000 National Cotton Variety Trials. After removing environmental variation, the model (which is based on genotype and genotype by environment interactions) explained 84% of the yield variation. The biplot made it is easy to see which varieties were best adapted to a particular location. The biplot further indicated that testing environments fell into three sector groups based upon vertex varieties. Cluster analysis based upon between location correlations identified five environmental clusters which is also a reduction from the currently utilized seven in the testing program. The calculation of an average tester coordinate indicated that two locations, St. Joseph, LA and Stoneville, MS were most representative of the average environment and were able to discriminate between varieties quite efficiently. The varieties closest to being ideal for both yield and stability across all of the testing locations were Stoneville 474 and Suregrow 125. The ability of GGEbiplot to analyze multi-location data and perform which-won-where queries should prove to be valuable in advancing plant breeding research.