This paper reports the results obtained in four experiments conducted at different localities of the State of São Paulo with the dwarf variety of castor beans IA-38 and designed to study the effect of nitrogen (Chilean nitrate), phosphorus (superphosphate) and potassium (potassium chloride), applied alone or in various combinations, under three plant spaeings: wide (1.50x1,20m), medium (1.00x0.90m) and close (1.00x0t45m). In three of the experiments, which were carried out under fairly favorable conditions, the average yields obtained with the wide spacing were lower than those obtained with the medium or close spacings, the difference betwen these two being very small. The three nutrients tested, principally nitrogen and potassium, greatly influenced the determination of the best spacing; while in their absence the advantage of increasing the density of plants was null or small, in their presence it became considerable. Likewise the responses to those nutrients, particularly to nitrogen and potassium, which were small or even negative with the wide spacing, increased substantially under the medium or close spacings. One of the experiments was carried out under poor conditions, principally because the rainfall was far below normal. In this case the yields were very low, the reduction of the spacing did not increase them and the responses to the nutrients tested, viz. phosphorus and potassium, decreased as the density of plants was increased. As the wide spacing used in the present experiments was much closer than those used in most of the previous experiments with castor beans, the authors assumed that the adoption of too wide spaeings have contributed appreciably to reduce the effect of the fertilizers then tested.