In this third paper on phosphorus fertilizers for cotton the authors report the results obtained in eight experiments carried out from 1949-50 to 1954-55, in which superphosphate was compared with Hyperphosphate and Serrote-phosphate (ground rock phosphates) on the basis of 0, 40 and 80 kilograms of P2O5 to the hectare in the presence of nitrogen and potash. The experiments were conducted for 2-5 yeors in the same plots and located on different soil types of the Planalto Paulista. In the overage of the six experiments located on soil which responded to phosphorus, with the 40-kilogram rate of application the yield increases due to Hyperphosphate and Serrote-phosphote were respectively half and one fourth of that caused by superphosphate; with the 80-kilogram rate, however, the response to Hyperphosphate was a little superior and that of Serrote-phosphate about equal to half the effect of superphosphate. The relative responses to the three phosphates were practically equal in the various soil types, but they changed appreciably in the course of the experiments. In the first year, while superphosphate increased considerably the yields, the responses to the other phosphates were very poor, even when they were applied at the 80-kilogram rate. However, starting from the second or third year the effects of the two rates of Hyperphosphate were a little superior to half the effects of the corresponding doses of superphosphate. Regarding Serrote-phosphate, at the 40-kilogram rate its effect was always very poor, but with the double rate it proved about half as efficient as superphosphate with the same amount of phosphorus.