Except at very low P rates, residual soil P (i.e., that measured by most extraction methods) continues to increase with annual additions of P fertilizer. Phosphorus retention mechanisms, however, are still much debated and difficult to prove or disprove. In an effort to determine whether the solubility of phosphate minerals controlled soil-solution P concentration in the field under some conditions, samples from the Ap of a Benndale sandy loam (Typic Paleudults) were collected from plots of a field experiment with different pH and P levels and analyzed in the laboratory. Several soil pH levels were first established (5.0–8.0); this was followed by annual additions for 7 yr of concentrated superphosphate at five rates varying from 0 to 392 kg P ha−1 yr−1 and cropping annually. Soil-solution composition was determined on samples taken 1, 3, and 5 yr after discontinuing P fertilization. Except for the “no P” treatment, solution P increased with increasing pH up to pH 5.8, then it decreased as pH increased, suggesting the accumulation of a basic phosphate mineral such as hydroxyapatite at about pH 5.8 and above. However, this was not matched by the ion activity product (IAP) for Ca5OH(PO4)3 or any other P mineral. For instance, at each P level, pCa5OH(PO4)3 decreased almost linearly with increasing pH so that each P level had one and only one pH at which pCa5OH(PO4)3 equalled 57.5, the accepted solubility product for hydroxyapatite. Thus, IAP's could not be used to predict, even indirectly, that solid-phase minerals were or were not controlling solution P. Sequential equilibration of selected soil samples with 0.01 M MgCl2, however, provided data that strongly supports the hypothesis that hydroxyapatite was present in a “high P, high pH” soil but not in a “low P, high pH” or a “high P, low pH” soil.