The authors bring together two strands of the empirical literature and analyze the geography of the regional economic performance of the states of the Brazilian Federation from 1939 to 1998. Using tools from spatial statistics, they examine the spatial dependence of regional per capita income in Brazil during the past six decades. They also examine the role of geography in explaining economic growth patterns using intradistribution dynamic tools based on Markov transition matrices and stochastic kernels in a discrete and a continuous framework. The analyses reveal the existence of two spatial clusters in Brazil, a low-income cluster in the northeast and a high-income cluster in the southeast. Moreover, the spatial pattern of economic growth in Brazilian states cannot be viewed without accounting for spatial spillovers. The authors show that states with relatively rich neighbors have a greater chance of being prosperous.