Soil phosphorus (P), which is potentially a risk for environmental contamination, is currently interpreted using soil P saturation in North America. Our objective was to assess the ratio of P to aluminum (Al) in the Mehlich-III (M-III) soil test to build P requirement models for corn and soybean. We analyzed 129 corn and 19 soybean P fertilizer trials. For corn, the (P/Al)M-III ratio improved soil fertility classification compared with PM-III alone. The critical PM-III value as determined by the Cate-Nelson procedure was found to be 31.5 mg PM-III kg-1, close to published values. The critical (P/Al)M-III ratios of 0.025 for > 300 g clay kg-1 soils and 0.040 for ≤ 300 g clay kg-1 soils differed significantly between the two soil groups. For (P/Al)M-III ratios above 0.214, there was no positive response to added P for all soils regardless of texture. Using published critical environmental (P/Al)M-III ratios of 0.076 for > 300 g clay kg-1soils and 0.131 for ≤ 300 g clay kg-1 soils as benchmarks values, agri-environmental P requirement models were built using conditional expectations of 50 to 80% of computed optimum P values within a soil class. A validation study supported the low critical (P/Al)M-III ratios and the 50% conditional expectation model except for a high carbon soil which was outside the application range of the models. However, banded P decreased corn yield at four validation sites although the model predicted positive response to P. Soybean did not respond to P except at extremely low fertility levels ((P/Al)M-III ≤ 0.02) and behaved as a P-mining crop even in low-P soils. Corn-soybean rotations can reduce soil P to low (P/Al)M-III ratios with minimal agronomic risk. Key words: Soil phosphorus saturation, Mehlich-III soil extraction method, soil fertility classification, soil texture, fertilizer P requirement model, corn, soybean