A study was made of the efficiency of five P fertilizer sources at three P205 levels (100, 200, 300 kg/ha) in their effect on the dry matter yield of Italian ryegrass Lolium multiflorum cv. 'Westerwool' (tetraploid), and also on the amounts of available Pin the soil and P removed by the plant. The foliage was cut five times with 50 day intervals between each cutting. The P sources used were ammonium polyphosphate, urea rock phosphate, ammonium-urea orthophosphate, magnesium ammonium potassium phosphate (Magamp) and triple superphosphate. These were tested in three soils with different phosphate fixing capacities: Tabasco, with a fixing capacity of 92 per cent and pH 4·5; Hidalgo, 90 per cent and pH 5·9; and Chapingo, 10 per cent and pH 7·0. In the oils with high fixing capacities (Hidalgo and Tabasco) Magamp, regarded as a controlled release fertilizer, gave the highest dry matter yields and supplied the greatest amount of available P as indicated by the P uptake of the crop. Urea rock phosphate gave the lowest dry matter yields, while among the other sources, ammonium polyphosphate, triple superphosphate and ammonium-urea orthophosphate there were no significant differences in yield. In the Chapingo soil which presented no problems of P fixation, all the sources showed a similar pattern of behaviour from the second cut onwards, but at the first cut the rapid release sources Such as triple superphosphate and ammonium-urea orthophosphate gave higher yields.