Although simulation models have been developed to describe P dynamics in cropped soils, they are designed for the plow layer, in which soil P is assumed to be homogeneously distributed. They are not suitable for soils cropped under no-till systems (NT), in which P distribution in the arable layer is strongly stratified. This work was undertaken to develop and test a two-dimensional distribution model (2D-CycP) for plant-available soil P in the arable layer under NT. To account for lateral and vertical P stratification, we developed a process-based P cycling model adapted to the NT context by partitioning the arable layer into elemental units, units occupying 5 × 10 or 10 x 10 cm sections of the soil profile. For each unit, the plant-available P is balanced by the annual P budget (inputs minus outputs). The main soil input variables of the 2D-CycP model are orthophosphate ion concentration (Cp) in the soil solution at the onset of NT, soil type-specific parameters to describe generalized kinetics of sorption-desorption, root biomass and surface density, and annual P budget. We evaluated the ability of the 2D-CycP model to predict plant-available P dynamics in soils cropped under corn-soybean rotation at different P doses (0, 17.5 and 35 kg P ha−1 applied once every 2 years) over several decades under NT. We used data collected in 2014 after 23-years of field experiment (Quebec, Canada). The 2D-CycP model reproduced patterns of multiannual dynamics of lateral and vertical Cp stratification across the arable layer in relation to P doses. Simulations agreed with Cp measurements when no P was applied. However, there were mismatches between measurements and simulations when P granules were assumed to have always been placed in the same position every fertilization year. Simulations were improved by accounting for more realistic agricultural field conditions such as uncertainty of P fertilizer placement position and soil mixing by earthworm bioturbation.