Yard long bean (Vigna unguilata L.) 777 was grown in Ultisols, which typically have low pH and high P-fixation, to determine the best correlation of soil extraction methods for soil P with yields, and to develop soil P response categories. The research was conducted at SANREM base camp in Hambaro Village, Nanggung, Bogor, Indonesia from April-August 2008. Treatments were arranged in a Split Plot Design with three replications. The main plots were treatments with soil P status of 0X, 1⁄4X, 1⁄2X, 3⁄4X and X, where is X = 1,590.5 kg SP-36 ha-1 (36% P2O5) applied once a month before planting. The subplots were P application rate of 0, 75, 150, 225 and 300 kg P2O5 ha-1. Yard long beans were planted in double rows per bed, 60 cm between rows and 25 cm within rows, 2 seeds per hole, with plot size of 1.5 m x 5 m. Coefficient correlation (r) of extraction reagents Olsen, Bray-1, HCl 25%, and Mechlich-1 were 0.772, 0.765, 0.755, and 0.732, respectively. Based on Olsen soil testing methods, soil response categories of very low, low, medium, and high were (ppm P2O5) ≤ 18.40, 18.40 < P < 117.27, 117.27 < P < 267.04, and ≥ 267.04 extracted-P, respectively. Based on Bray-1 soil testing methods, soil response categories for low, medium, and high were ≤ 87.81, 87.812O5), respectively. Fertilizer recommendation based on the Olsen soil test for low response category was 185.75 kg P2O5 ha-1, and for the medium soil category was 175.97 kg P2O5 ha-1. The Bray-1 soil test for the low response category was 184.31 kg P2O5 ha-1, and for the medium soil category was 161.39 kg P2O5 ha-1. Keywords: fertilizer recommendation, calibration, phosphorus, yard long bean