This paper reports the results obtained with upland rice in two long-term experiments located at Campinas, on "terra-roxa-misturada" soil, in which various sources of phosphorus and potassium fertilizers were compared. In both trials rice occupied the plots after 4 crops of corn, 2 of cotton, and 4 of small grains, each receiving the fertilizers indicated below. For upland rice, the yields were relatively high in the NPK-plots. In the phosphorus experiment (with superphosphate, Thomas slag, Rhenaniaphosphate, and two kinds of bone meal), the response to 60 kilograms of P2O5 per hectare in the presence of NK averaged +48%. superphosphate being slightly superior to the other phosphates. Although very good, this response was much smaller than those obtained in the corn and principally in the cotton crops. Of the potassium sources (kainit, potassium-magnesium sulfate, potassium sulfate and potassium chloride), applied at the fate of 70 kilograms of K2O to the hectare in the presence of NP, the chloride, which was the best, increased the yield of only 11%. This small response could be attributed to the use of sodium nitrate as nitrogen source. However, under the same conditions the effect of potassium was good on the corn and very high on the cotton crops.