In many states, Morgan (M) or Mehlich-3 (M3) extractions form the basis for phosphorus (P) fertility recommendations and P runoff risk assessment.  Currently, the impacts which the timing of sampling after P application, the source of applied P, and other inherent soil chemical properties (extractable Al and Fe) have on agronomic soil test P and P release to the environment, as estimated by 0.01 M CaCl₂ extractable P, are not well-understood.  Two laboratory incubation studies were conducted to determine i) changes in M-P, M3-P, and CaCl₂-P over time following the application of CaH₂PO₄; ii) the efficiency of liquid dairy manure, NH₄HPO₄, and CaH₂PO₄ in raising M-P and M3-P levels; and iii) the degree to which reactive Al and Fe alter the efficiency of applied P.  Experiments were conducted on 28 New York soils of non-calcareous origin, which were incubated in the dark at 23^oC and field capacity (0.33 Bar). The amount of P decreased over the 60 day incubation for all soil tests, but a greater proportion of P was lost from the more labile pools (CaCl₂-P > M-P > M3-P).  Changes in soil test P per unit P added (ΔP-M and ΔP-M3) were significantly affected by P source, with NH₄HPO₄ more efficiently raising P levels than either CaH₂PO₄ or liquid dairy manure. Extractable Al was inversely related to ΔP-M and ΔP-M3 and followed exponential decay functions. Extractable Al was positively and linearly related to the amount of P required to raise M-P and M3-P levels 1 mg kg^-1, with Morgan extractable Al as the most accurate predictor of soil test P increase. Our results suggest that components of the NY P runoff index may need to include guidance for the time of sampling and account for the improved sorption of P by soils inherently high in Al.