Studies were conducted in the States of Goiás and Tocantins, Brazil, to develop a sampling program for rice stink bug Oebalus spp. (Heteroptera: Pentatomidae) in rice. Initially, a sampling survey to describe the spatial and the probability distributions of adult stink bugs was conducted using sweep net taking 10-sweep samples per point. A total of 29 fields (10 to 20 ha) were sampled over three years. Fourteen fields were randomly sampled between 1999 and 2002 and fifteen fields were sampled in 2004 following a regular grid, keeping a minimum distance or 10 m between sampling locations. All fields were sampled when more than 50% of the panicles were exposed and the spikelets were in the beginning of milky stage. Semivariograms showed sign of aggregation in only two fields, even though the presence of macroscale trends indicated that stink bug densities may be greater in the sites infested with weeds (Echinocloa spp) which serve as alternative hosts, and near edges of adjacent to harvested fields. A sequential sampling plan based on negative binomial distribution was developed and implemented in 18 rice fields in 2007 growing season. The average sample number required to classify rice stink bug population as above 5 or below 3 ranged from 5 to 25, with samples taken at distance of 100 m. The results presented in this study support the hypothesis that the use of this sampling plan will be efficient to help growers and contribute to reduce the number of insecticide applications.