The residual value of phosphorus (P) applied at 0 to 320 kg P/ha as superphosphate 9 to 12 months earlier was measured using yields of carrots(Daucus carotaL. cv. Western Red). The effectiveness of the previously‐ applied P was determined relative to the effectiveness of freshly‐applied P on an acutely P‐deficient, newly cleared Karrakatta sand in experiments over three years. Yield response was significantly (P < 0.01) higher to freshly‐applied P than residual P at all rates except 320 kg/ha. The relative effectiveness of residual P increased from 46% of freshly‐applied P at 20 kg/ha to 95% at 320 kg/ha. Bicarbonate‐soluble P extracted from the soil was determined on the residual P site in each experiment. These soil test P values were related to yield. The critical soil test values required for 95% or 99% of maximum yield were 45 and 60 μg/g, respectively. The rate of freshly‐applied P required for 95% or 99% of maximum yield decreased from 118 and 183 kg P/ha at ≤ 5 μg/g soil test to 0 at the critical values. High levels of residual P (320 kg P/ha) resulted in high soil test values (≥ 60 μg/g) and were as effective as freshly‐applied P. Results of this work showed soil testing could be used as part of the P management of carrots on sandy soils. This has twin benefits of reducing fertiliser costs to the growers and reducing the leaching of fertiliser P into waterways thereby reducing eutrophication.