The advantage of split application of potassium chloride in cotton was studied in ten field experiments, conducted at different localities during the period of 1975 to 1981. In such cases, doses of 0, 60 and 120kg/ha of K2O, were placed in a fillet beside and below the seeds, at planting. In other treatments 1/3. 1/2, 2/3 or the total quantity of the highest level of potassium fertilizer was applied together with nitrogen, just after thinning, in a side-dressing operation, and covered slightly by subsequent cultivation. When all the K was applied at planting, linear effect on cotton yield appeared only in 10% of the experiments. This proportion increases to 40% when split applications were made. However, in combined analysis of all tests no significant difference was found between the two methods. Based in a study of simple correlations between cotton productivity and potassium application, expressed as the control treatment's relative yield and indices of soil analysis ike 1/K+ and (Ca2+ + Mg2+)/K+, three groups of experiments were established, according to different expected responses. Considering the group of high expected responses [K < 0.08 meq and (Ca2+ + Mg2+)/K+ > 43], parcelled application resulted in increases of cotton yield when compared with the normal method of application. Similar tendency was observed in the group of intermediate expected responses [0.8-0.24 meq of K+ and 20-43 of (Ca2+ + Mg2+)/K+]. In potassium deficient soils split application of potassium chloride can be recommended, 1/2 or 2/3 of the total level furnished at planting and the remainder just after thinning, by side-dressing, mixed with nitrogenous fertilizer and followed by cultivation.