this paper reports the data obtained in the third period of an experiment which was started in 1954, to study the recovering for sugar cane culture of an original low fertility and impoverished soil (humic red-yellow latosol) . at the same time with the results of the two harvestings of this third planting one discusses the soil productivity and fertility variations which had occurred up to the first planting. in a randomized blocks design eight treatments with four replications were compared. they are as follows: a) control; b) p + lime + leguminous cover crop; c) pk + cover crop; d) pk + lime + cover crop; e) npk; f) npk ++ lime; g) npk + cover crop; h) npk + lime + cover crop. in this period the cover crop still used was the sun hemp (crotalaria junceal.) and the sugar cane variety planted was cb 41/76. in the planting time the fertilization applied was at the rates of 90-80-120 kg per hectare of n, p2o5 and k2o respectively and as ammonium sulfate, plain superphosphate and potassium chloride. after the first harvesting other applications of nitrogen and potassium were made, at the rates of 60 kg and 90 kg per hectare respectively, using the same sources of fertilizers in top-dressing application. the yields of sugar cane obtained in this period showed appreciable effects due to lime and potassium and a considerable increase in soil productivity as compared with the data corresponding to the other ones. this result is attributed to the soil fertility increase, on account of favorable variations in soil acidity and ca, mg and po4 levels, as shown by chemical analyses. in spite of the leguminous cover crop usage and chemical fertilizations, no increase in carbon, nitrogen and potassium soil levels were detected.