1. A field experiment on an acid soil long in arable cropping at Rothamsted measured residual effects of superphosphate, applied at several rates, and Gafsa rock phosphate applied at a single rate. Residues were valued after 1 year with barley, and after 2 years with rye-grass. In each year, the value of the residues was related to yields from fresh superphosphate dressings taken as standards. ‘Percentage fresh superphosphate equivalents’ of the residues were determined using yields and P uptakes of both crops. An isotope dilution method with 32P was also used with barley. The experiment was continued for 2 further years without applying phosphate fertilizers, to measure yields and P uptakes from the residues and yields and P uptakes from rock phosphate relative to superphosphate.2. ‘Percentage fresh superphosphate equivalents’ of residues of superphosphate for barley varied with the growth of the crop and with the method used to calculate them. They were 21–24%6 weeks after sowing, measured from yield, P uptake or isotope dilution. At harvest, they were 26% from yield, 43% from P uptake and 49% by isotope dilution. ‘Percentage fresh superphosphate equivalents’ 6 weeks after sowing increased with the rate of superphosphate but at harvest were independent of rate. ‘Percentage superphosphate equivalents’ of Gafsa rock phosphate were much smaller; they increased from 2–3% 6 weeks after sowing to 7–12% at harvest.3. With rye-grass ‘percentage fresh superphosphate equivalents’ of residues of superphosphate applied either 1 or 2 years previously also varied with growth. All methods of valuation showed that residues were about twice as effective after 1 year as after 2 years in the soil. ‘Percentage fresh superphosphate equivalents’ derived from P uptake remained constant during growth and were 37–38% for superphosphate applied 1 year before and 18–20% for superphosphate applied 2 years before. Values derived from rye-grass yields decreased during growth from 76 to 45% (1-year residues) and from 38 to 21% (2-year residues).4. After cropping with barley for 2 more years, there was little difference between yields or P uptakes from equal amounts of superphosphate whether applied 3 or 4 years previously. Residues from rock phosphate were almost equivalent to those from superphosphate after 3–4 years.5. Apparent recoveries of superphosphate, as percentages of the amount applied, decreased with rate. The apparent recovery of P by crops in 5 years was about 21% from superphosphate and about 10% from rock phosphate when both were applied at 3·0 owt. P2O5 per acre.