Understanding the forms of phosphorus present in soil is important for elucidating its accumulation, migration, transformation, and bioavailability in the environment. At present, however, studies on the spatial variability of different forms of phosphorus in soil across large-scale transects are rare. In the present study, seven zonal forest soils from sites across different climatic zones were collected along a latitudinal transect in eastern China. The soils used included brown coniferous forest soil from the cold temperate zone, dark brown soil from the middle temperate zone, brown soil from the warm temperate zone, yellow brown soil from the northern subtropical zone, yellow soil from the middle subtropical zone, lateritic red soil from the southern subtropical zone, and latosol from the tropical zone. A chemical extraction method was combined with solution phosphorus-31 nuclear magnetic resonance (31P NMR) spectroscopy to analyze the phosphorus forms present in the soils and determine their relationships with other soil properties. The concentrations of total phosphorus, available phosphorus, inorganic phosphorus, and organic phosphorus in the tested soils ranged from 179.8 to 825.2, 2.41 to 15.3, 92.6 to 351.2, and 14.7 to 474.4 mg·kg-1, respectively. The concentrations of four organic phosphorus components (i. e., active, moderately active, moderately stable, and highly stable organic phosphorus) obtained through continuous chemical extraction were 1.38～30.9, 8.63～213.7, 3.01～32.2, and 1.73～199.2 mg·kg-1, respectively. According to solution 31P NMR spectra, both inorganic (i. e., orthophosphate and pyrophosphate) and organic (i. e., phosphomonoester, phosphodiester, and phosphonate) forms of phosphorus were identified in the test soils. Moreover, neo-inositol hexakisphosphate, D-chiro-inositol hexakisphosphate, RNA mononucleotides, α-glycerophosphate, myo-inositol hexakisphosphate, β-glycerophosphate and scyllo-inositol hexakisphosphate in phosphomonoesters and deoxyribonucleic acid in phosphodiesters were also identified. In all the tested soils, polyphosphate was not detected. Phosphonate was not detected in the soils except in brown coniferous forest soil and dark brown soil, whereas deoxyribonucleic acid was not detected in lateritic red soil. Inorganic phosphorus was dominated by orthophosphate, while organic phosphorus was dominated by phosphomonoester. In general, regardless of whether chemical extraction or solution 31P NMR spectroscopy, the concentrations of total phosphorus, available phosphorus, inorganic phosphorus and organic phosphorus and its fractions tended to decrease from brown coniferous forest soil in the cold temperate zone to latosol in the tropical zone. There was a correlation between phosphorus forms identified using solution 31P NMR spectroscopy and those identified using chemical extraction method. Orthophosphate was most closely related to active organic phosphorus; phosphomonoester and phosphonate were most closely related to moderately active organic phosphorus; and pyrophosphate and phosphodiester were most closely related to moderately stable organic phosphorus. 31P NMR spectroscopy is a more effective method than chemical extraction to understand the spatial variability in soil phosphorus at a detailed molecular level.