Eighteen 3³ factorial experiments using 0, 80, 160 kg/ha (kilograms to the hectare) of N and K2O, and 0, 90, 180 kg/ha of P2O5 were conducted in 1957-58 in farms belonging to the sugar factory Itaiquara, located at Tapinitiba, State of São Paulo. The areas studied, all of the "massapê-salmourão" type of soil, had been cropped during different periods and, when fertilized, received a mixture containing almost exclusively phosphorus. The effect of phosphorus was significant only in six of the 18 experiments. The average responses to it readied +40.3 and +43.1 t/ha (metric tons of cane per hectare) in two trials located on areas which had received one phosphorus application in the previous crops, but varied between -3.5 and +12.5 t/ha in the 16 areas that had received two to five such applications. Contrarily to that of phosphorus, the effect of potassium was significant in all but one trial and the average responses to the rates of 80 and 160 kg/ha of K2O corresponded to +12.9 and +23.5 t/ha, respectively. The effect of nitrogen was generally unsatisfactory and significant only in six experiments. Even in the presence of potassium, which increased the need for it, the average response to its larger dose was +9.8 t/ha. While the average yield was 90.6 t/ha in the unfertilized plots and 101.4 t/ha in those exclusively fertilized with 90 kg/ha of P2O5, (a treatment similar to that previously used by the farmer), that of the plots which received the larger doses of the three elements reached 135.8 t/ha. Most of the areas studied were cropped for long time and fertilized almost exclusively with phosphorus. These facts explain the high responses to potassium and the small effect of phosphorus in the greatest part of the trials. The information available are not sufficient to explain why, in such conditions, the responses to nitrogen were not higher.