Agricultural productivity is characterized by location patterns, represented by the so-called spatial effects, namely, spatial heterogeneity and spatial dependence. This paper analyzes the spatial effects which exert influence on agricultural land productivity at three points in time – 1991, 1997 and 2003 – at the microregional level, using exploratory spatial data analysis (ESDA). The main finding shows that agricultural productivity is increasingly self-correlated in space, over time. Two high-high (HH) clusters are identified consistently: one of them is located in the State  f São Paulo and parts of the Center-West region, while the other cluster (HH) lies in the Northeastern Coast.