Size data is the basic information in fishery stock assessment and management. The spatio-temporal distribution of length-frequency derived from one fisheries dependent data can be used to estimate the distribution of its stock and recruitment. Multivariate regression tree (MRT) is a data-mining methodology that can analyze the relationships between multispecies data and environmental characteristics. In this study, based on the dressed weight data in the logbooks collected by Tuna Technology Group of China Distant-water Fisheries Association from December 2007 to December 2009, Multivariate Regression Tree and Geographical Information System (GIS) were applied to analyze spatio distribution and seasonal variation of bigeye tuna size in the Central Atlantic Ocean. The result showed that the large-sized bigeye tuna mainly distributed in 7.5°N-15°N, 17.5°W-45°W area; Medium-sized bigeye tuna mainly distributed in 12.5°S-5°N, 17.5°W-45°W area; Small-sized individuals mainly distributed in 7.5°S-5°N,5°W-17.5°W area. The results of K-S test showed that size distribution had seasonal differences apparently; Kolmogorov-Smirnov test proved that there were significant differences in size distribution among spatial areas. Quarter 1 and 2 appeared relatively homogeneous, and quarter 3 and 4 appeared relatively homogeneous.