The effects of P and K on grapevines (Vitis vinifera L.) var. Muscat Rosada were evaluated four years, considering the same levels of N. The experiment was conducted at the Vicuña Experimental Station (30° S; 70°44´ W) of the Instituto de Investigaciones Agropecuarias (INIA). The soil is alluvial antropic miscellaneous (Entisols). Three fertilization treatments were established: 1) N 160 + P2O5 200 + K2O 300; 2) N 160 + P2O5 0 + K2O 300; and 3) N 160 + P2O5 200 + K2O 0. At the beginning of the experiment, grape vines received all the phosphate fertilizers, 200 kg ha-1 of P2O5 as triple superphosphate in Treatments 1 and 3. Nitrogen was applied as urea for four years. The applications were initiated when the vine buds measured 20 cm. The K was applied to the soils of Treatments 1 and 2, at a dosage of 200 kg ha-1 K2O, using potassium sulphate. Between berry set and 30 d before harvest, 100 kg ha-1 K2O were applied as potassium sulphate by fertigation. Significant effects of the P fertilization were observed by the second year, with increased cluster numbers per plant. K also increased grape fruit yield by the third year. This response was obtained with less than 5 mg kg-1 of P available in the soil and less than 145 mg kg-1 of exchangeable K in the soil. The initially low content of available K and P suggests a high probability of response to the application of both elements.