Considerable amounts of residual fertilizer phosphorus (P) have accumulated in the agricultural soils of Finland since the 1960s, and the P fertilization recommendations have been lowered. It is unknown how much P intensively managed silage grass can obtain from the accumulated reserves without a loss of yield. In two field experiments on sandy loam conducted in 2003–2020, four consecutive grass (70% timothy, 30% fescue) rotations were performed (four or five years each, including the establishment year). The grass received mineral P fertilizers (PF; 16 kg P ha−1 year−1), cattle slurry (PS; 11 kg P ha−1 year−1) or no P (P0). The organic P (Po) and inorganic P (Pi) pools in 2003 and 2020 samples were determined following the Hedley procedure using H2O, NaHCO3, NaOH, and HCl as sequential extractants. Soil test P (STP) was monitored annually using ammonium acetate extraction. The results showed that the cumulative P balance (P0: −344– −412 kg ha−1; PF and PS: −101– −198 kg ha−1) was highly negative, resulting in declining STP. Still, after 18 years, the grass showed no consistent yield response to P fertilization. The most significant Pi decline occurred in the Pi–NaHCO3 (~30%) and Pi–NaOH (~50%) pools, while the changes in Po were negligible. This study and international comparisons, Mehlich‐3, degree of P saturation and the result of Hedley in other studies, suggest that these soils, initially above the critical STP level, contain plenty of legacy P and can provide perennial grass with sufficient P for a long time.