In this paper we analyze the Indirect Land Use Change (ILUC) effects of ethanol production expansion in Brazil through the use of an inter-regional, bottom-up, dynamic general equilibrium model calibrated with the 2005 Brazilian I-O table. A new methodology to deal with ILUC effects is developed, using a transition matrix of land uses calibrated with Agricultural Censuses data. Agriculture and land use are modeled separately in each of 15 Brazilian regions with different agricultural mix. This regional detail captures a good deal of the differences in soil, climate and history that cause particular land to be used for particular purposes. Brazilian land area data distinguish three broad types of agricultural land use, Crop, Pasture, and Plantation Forestry. Between one year and the next the model allows land to move between those categories, or for Unused land to convert to one of these three, driven initially by the transition matrix, changing land supply for agriculture between years. The transition matrix shows Markov probabilities that a particular hectare of land used in one year for some use would be in an other use next period. These probabilities are modified endogenously in the model according to the average unit rentals of each land type in each region. A simulation with ethanol expansion scenario is performed for year 2020, in which land supply is allowed to increase only in states located in the agricultural frontier. Results show that the ILUC effects of ethanol expansion are of the order of 0.14 hectare of new land coming from previously unused land for each new hectare of sugar cane. This value is higher than values found in the Brazilian literature. ILUC effects for pastures are around 0.47. Finally, regional differences in sugarcane productivity are found to be important elements in ILUC effects of sugar cane expansion.