Accurate fertilizer recommendations require numerous site-years of data to correlate soil nutrient availability indices to crop yield response and calibrate the fertilizer rates associated with designated soil-test levels. The objective of this project is to conduct phosphorus (P) and potassium (K) correlation and calibration trials with rice, soybean, and winter wheat on silt-loam soils in eastern Arkansas. Soil chemical property and crop yield data were assimilated from 14 site-years for winter wheat response to P, 9 site-years for soybean response to P, 11 site-years for soybean response to K, and 16 site-years for rice response to K. Relative yield data for the unfertilized controls were regressed against the Mehlich-3 nutrient concentrations. The relationship between Mehlich-3 K and relative soybean yield was significant (P = 0.0002), nonlinear, and Mehlich-3 K explained 88% of the variability in relative soybean yields. Predicted relative soybean yields of 90 and 95% were produced by Mehlich-3 K concentrations of 100 and 115 mg K/kg, respectively. Correlations for other relative yields and nutrients were linear and significant (P <0.10), but explained only 23 (rice-K), 35 (wheat-P), and 37% (soybean-P) of the variation in relative yields. For wheat, relative yields <90% were produced at five sites which also had <25 mg P/kg. Soils with >25 mg P/kg never produced relative wheat yields <92%. For soybean, relative yields <90% were produced at only two sites which had <20 mg P/kg. However, three other sites with <20 mg P/kg had relative soybean yields >92%. For rice, relative yields <90% were produced at eight sites with Mehlich-3 K <100 mg K/kg. Five other sites also had <100 mg K/kg and produced relative rice yields >90%. Fertilization trials will be continued to build the correlation and calibration databases for each crop.