Because carbon dioxide (CO2) concentration is rising, increases in plant biomass and productivity of terrestrial ecosystems are expected. However, phosphorus (P) unavailability may disable any potential enhanced growth of plants in forest ecosystems. In response to P scarcity under elevated CO2, trees may mine deeper the soil to take up more nutrients. In this scope, the ability of deep horizons of forest soils to supply available P to the trees has to be evaluated. The main objective of the present study was to quantify the relative contribution of topsoil horizons and deep horizons to P availability through processes governed by the activity of soil micro-organisms. Since soil properties vary with soil depth, one can therefore assume that the role of microbial processes governing P availability differs between soil layers. More specifically, our initial hypothesis was that deeper soil horizons could substantially contribute to total plant available P in forested ecosystems and that such contribution of deep horizons differs among sites (due to contrasting soil properties). To test this hypothesis, we quantified microbial P and mineralization of P in ‘dead’ soil organic matter to a depth of 120 cm in forest soils contrasting in soil organic matter, soil moisture and aluminum (Al) and iron (Fe) oxides. We also quantified microbiological activity and acid phosphomonoesterase activity. Results showed that the role of microbial processes generally decreases with increasing soil depth. However, the relative contribution of surface (litter and 0–30 cm) and deep (30–120 cm) soil layers to the stocks of available P through microbial processes (51–62 kg P ha−1) are affected by several soil properties, and the contribution of deep soil layers to these stocks vary between sites (from 29 to 59%). This shows that subsoils should be taken into account when studying the microbial processes governing P availability in forest ecosystems. For the studied soils, microbial P and mineralization of P in ‘dead’ soil organic matter particularly depended on soil organic matter content, soil moisture and, to a minor extent, Al oxides. High Al oxide contents in some sites or in deep soil layers probably result in the stabilization of soil organic compounds thus reducing microbiological activity and mineralization rates. The mineralization process in the litter also appeared to be P-limited and depended on the C:P ratio of soil organic matter. Thus, this study highlighted the effects of soil depth and soil properties on the microbial processes governing P availability in the forest spodosols.