The aim was to assess the ability of bicarbonate‐extractable P (Olsen P) to estimate total plant‐available P (TPAP) in reclaimed marsh soils (Aeric Endoaquepts) which differed widely in P buffering capacity (PBC). Total plant‐available P was estimated as the cumulative P uptake for a final concentration of 0.02 mg P/L in the soil solution which is the typical P requirement for field crops. The Olsen P estimated for that concentration was adopted as the critical level for crop production. We found that TPAP was better predicted by anion exchange resin‐extractable P (AER‐P) (65% of variance accounted for) than by Olsen P, probably because the effectiveness of the AER depends on the soil P buffering capacity, a factor that greatly influences the availability of P to plants. The critical Olsen P level was found to depend on those soil properties affecting the relationship between sorbed P and P in soil solution, viz. the P buffering capacity of soil, the Na/Ca mole ratio in the 1:1 soil:water extract, which explained 63 and 84% of the variance in the critical level, respectively, and the affinity of the sorbing surfaces for P. These properties must be taken into account when using Olsen P as the P index for fertilizer management.