Nine experiments with potatoes were conducted in 1962 on the lowlands of the Paraiba Valley, State of São Paulo, where they are planted during the dry period (winter) and irrigated by the subsurface method. In those experiment were studied, in a 3³ factorial scheme, increasing rates of application of nitrogen, phosphorus, and potassium. Averaging the seven experiments that could be analysed jointly, the responses due to N, P and K were positive, but only those of N and P were significant. The interactions N x P, N x K, and P x K were insignificant; however, the responses to N, P, and K showed to be much higher where each element was used in the presence of the other two. Under such conditions, the mean responses to N, P. and K corresponded to +37, +28, and +11 per cent, respectively. One of the two trials excluded from the joint analysis had low and irregular stands. In the other, however, the three nutrients significantly increased the yields, the average increase due to P having reached +354 per cent. While this trial was located on virgin soil, the other eight were conducted in areas fertilized with NPK in the previous crops.