Sequential chemical extraction has been widely used to study soil phosphorus (P) dynamics and inform nutrient management, but its efficacy for assigning P into biologically meaningful pools remains unknown. Here, we evaluated the accuracy of the modified Hedley extraction scheme using P K-edge X-ray absorption near-edge structure (XANES) spectroscopy for nine carbonate-free soil samples with diverse chemical and mineralogical properties resulting from different degrees of soil development. For most samples, the extraction markedly overestimated the pool size of calcium-bound P (Ca-P, extracted by 1 M HCl) due to (1) P redistribution during the alkaline extractions (0.5 M NaHCO3 and then 0.1 M NaOH), creating new Ca-P via formation of Ca phosphates between NaOH-desorbed phosphate and exchangeable Ca2+ and/or (2) dissolution of poorly crystalline Fe and Al oxides by 1 M HCl, releasing P occluded by these oxides into solution. The first mechanism may occur in soils rich in well-crystallized minerals and exchangeable Ca2+ regardless of the presence or absence of CaCO3, whereas the second mechanism likely operates in soils rich in poorly crystalline Fe and Al minerals. The overestimation of Ca-P simultaneously caused underestimation of the pools extracted by the alkaline solutions. Our findings identify key edaphic parameters that remarkably influenced the extractions, which will strengthen our understanding of soil P dynamics using this widely accepted procedure.