In this paper are reported the results obtained in six experiments comparing various potash sources as fertilizers for cotton. The experiments were conducted for some years in the same plots and the potash fertilizers were applied annually in the presence of nitrogen (Chilean nitrate) and phosphorus (superphosphate or bone meal). In five of the experiments, located at different soil types of the Planalto Paulista, potassium chloride and potassium sulfate gave similar results. One of these experiments, on "terra-roxa-misturada" soil highly deficient in potassium, had also treatments with kainit and potassium-magnesium sulfates, their effects being equal to that of potassium chloride. In the some experiment, magnesium sulfate applied in quantity equivalent to that contained in potassium-magnesium sulfates, and magnesium sulfate plus sodium chloride, in quantities equivalent to those contained in kainit, had practically no effect on the yield. In another experiment, on "terra-roxa-misturada" soil, leucite was compared with potassium chloride. While the response to the latter was very high, leucite induced practically no response. Severe symptoms of "potash hunger" were observed in both experiments conducted on the "terra-roxa-misturada" soil. The plants of the plots without potash lost their leaves prematurely. As a result, their bolls opened also prematurely and were considerably smaller than those of the plants well supplied with potash.