This paper discusses data obtained from a long-term N, P, K field experiment with maize (ZeamaysL.) on a Normandien clay loam (Plinthic Paleudult) near Newcastle, Natal. After eight seasons plots were split, sub-plots being allowed either to rundown or continuing to receive annual P applications. Soil samples were obtained shortly after fertilization each season and P soil tests were related to final grain yield. The results obtained indicate that although yield maxima varied from 3 000 kg ha−1in the dry 1981/82 season to over 6 500 kg ha−1in the more favourable 1983/84 season, there was no meaningful difference in the optimum P soil test using either the Ambic, modified Truog or Bray 1 extractants. Furthermore, the relationship between yield and soil test was identical for residual P subplots and those which received annual applications of fertilizer P.