During the years 1948, 1949 and 1951 a disease occurred in the cotton crops of the state of S. Paulo Brazil (S. Am.), which caused a severe drop in yields. The abnormality was characterized by a typical reddish - purple color of the leaves, being by this reason, called "vermelhão", that is, reddening of the cotton plant. The disease was associated with a dry season. Among the several hypotheses raised to explain the causes of the disease were: insect attack, potassium deficiency - where from the name "potash hunger" was also given -, and magnesium deficiency: In order to study the problem the Department of Agricultural Chemistry of the College of Agriculture of the University of São Paulo, at Piracicaba, carried out a series of experiments as follows: 1. pot experiments in which soil of one of the affected regions was used ("terra roxa", a red-brownish soil derived from basalt); 2. pot-soil experiments varying the moisture supplied; 3. sand culture experiments omitting certain elements from the nutrient solutions; 4. field plot experiments, conducted on a sandy soil; three different varieties were employed: Texas, Express, and I.A. 817; magnesium was applied either as sulfate or dolomitic limestone. All the experiments were completed with suitable chemical analyses. The results can be summarized as follows: 1. in the first trial, the not properly manured pots (minus Mg), symptoms were registered which were similar to the symptoms observed in the field; it was possible to establish some differences among three different types of reddening: due to lack of K in the mixed fertilizers used, the characteristic cotton rust made its appearance, the red color in the leaves of the minus Mg plants was all alike that described in the current literature as a symptom of Mg-deficiency; in all the treatments ocurred a yellow-reddish color in the leaves associated with the latest stages of maturity; 2. in the second experiment it was verified that when the plants in the pots with soil were kept 75 per cent of the water holding capacity, no symptom of deficiency showed up; was true even for the plants not receiving neither K nor Mg; however, plants supplied with only 25 per cent of the water holding capacity showed, respectively, cotton rust in the minus K treatment and the red purplish color in the minus Mg series; 3. the sand culture experiment confirmed lack of Mg as the cause of "vermelhão", being potash deficiency the responsible for cotton rust; 4. in the field experiment, variety LA. 817 revealed to be the most sensitive to "vermelhão" when Mg was omitted from the fertilizers; symptoms of K deficiency appeared when no K was supplied; both magnesium sulfate and dolomitic limestone proved to be equally effective in the control of "vermelhão"; 5. the analyses of material collected both in the field as well in the pots revealed that leaf petiole in the most reliable part to indicate the K and Mg status of the plant; the variation in Mg content suffered by the plants showing different stages of "vermelhão was, quantitatively, at least as large as that in K content, however when one deals with K deficient plants, that is, plants showing the typical rust, no variation occurred in the Mg content, whereas K in the dry mater dropped from more than 1 per cent to less than half per cent. Then, the following general conclusions can be drawn: 1. Mg deficiency is the cause of "vermelhão" of cotton crops; 2. K deficiency also occurred, but in a lesser degree; 3. the climate conditions - especially the lack of rain influenced the soil dynamic of K, and especially Mg, bringing a severe reduction in their assimilability; 4. the "vermelhão" disease can be easily controlled upon additions either of magnesium sulfate or dolomitic limestone.