Historically high N prices have increased farmer interest in economic optimum N rates for rice (Oryza sativa L.). This study uses five to 8 yr of rice variety by N data from four research locations in Arkansas varying by soil series Perry clay (Vertic Haplaquepts); Sharkey clay (Vertic Haplaquepts); DeWitt silt loam (Typic Albaqualfs); and Hillemann silt loam (Thermic, Albic, Glossic Natraqualfs) to estimate rice yield response to N functions for five common functional forms quadratic, quadratic-plateau, square root, linear-plateau, and Mitscherlich. Economic optimum N rates (EONs) are calculated by location and functional form across years using both nonlinear optimization and a numerical method proposed for making N rate guidelines in the Corn Belt of the United States. Statistical analysis indicated the quadratic fits the yield data best at three of four locations and performs as well as either the linear-plateau or quadratic-plateau at the fourth location. The numerical method produces approximately the same EON rates (within ±1 kg ha–1) as nonlinear optimization. Economic optimum N rates varied by research location and are as follows assuming average 2006–2008 rice price and N cost data and annual quadratic functions: 150 kg N ha–1 at the DeWitt silt loam location; 122 kg N ha–1 at the Hillemann silt loam location; 181 kg N ha–1 at the Sharkey clay location; and 187 kg N ha–1 at the Perry clay location.