Gross primary production (GPP) is one of the most important characteristics of an ecosystem. At present, no empirically based method to estimate GPP is available, other than measurements of net CO2 exchange and calculations of respiration. Data sets from continuous CO2 flux measurements in a number of ecosystems (Ameriflux, AgriFlux, etc.) for the first time provide an opportunity to obtain empirically based estimates of GPP. In this paper, using the results of CO2 flux tower measurements during the 1997 season at four sites in Oklahoma (tallgrass prairie, mixed prairie, pasture, and winter wheat crop), we describe a method to evaluate the average daytime rate of ecosystem respiration, Rd, by estimation of the respiration term of the nonrectangular hyperbolic model of the ecosystem-scale light-response curve. Comparison of these predicted daytime respiration rates with directly measured corresponding nighttime values, Rn, after appropriate length of the night and temperature correction, demonstrated close linear relationship, with 0.82 ≤ R2 ≤ 0.98 for weekly averaged fluxes. Daily gross primary productivity, Pg, can be calculated as Pg = Pd + Rd, where Pd is the daytime integral of the net ecosystem CO2 exchange, obtained directly from measurements. Annual GPP for the sites, obtained as the sum of Pg over the whole period with Pg > 0 were: tallgrass prairie, 5223 g CO2 m-2; winter wheat, 2853 g CO2 m-2; mixed prairie, 3037 g CO2 m-2; and pasture, 2333 g CO2 m-2. These values are in agreement with published GPP estimates for nonforest terrestrial ecosystems.