Drought stresses arise when the combination of rainfall and soil water supply are insufficient to meet the transpiration needs of the crop. In the Cerrado region of Goiás state, Brazil, summer rainfall is typically greater than 1000 mm. However, drought stress can occur during rain-free periods of only 1-3 weeks, since roots are frequently restricted to shallow depths due to Al-induced acidity in deeper soil layers. If these droughts are frequent, then plant breeding programs need to consider how to develop suitable germplasm for the target population of environments (TPE). A crop simulation model was used to determine patterns of drought stress for 12 locations and >30 environments (6 years × 5-6 planting dates) for short and medium duration rice crops (planted in early summer), and for maize grown either as a 1st or 2nd crop in the summer cycle. Regression analysis of the simulations confirmed the greater yield impact in both crops of drought stress (quantified as the ratio of water-limited to potential transpiration) when it occurred around the time of flowering and early grain-filling. For rice, mild mid-season droughts occurred 40-60% of the time in virgin (0.4 m deep for rice or 0.5 m for maize) soils and improved (0.8 m for rice or 1.0 m for maize) soils, with a yield reduction of <30%. More severe reproductive and grain-filling stress (yield reductions of 50% for rice to 90% for maize) occurred less frequently in rice (<30% of time) and 1st maize crop (< 10% of time). The 2nd maize crop experienced the greatest proportion (75-90%) of drought stresses that reduced yield to <50% of potential, with most of these occasions associated with later planting. The rice breeding station (CNPAF) experiences the same pattern of different drought types as for the TPE, and is largely suitable for early-stage selection of adapted germplasm based on yield potential. However, selection for virgin soil types could be augmented by evaluation on some less-improved