With the purpose of improving nutrition of the grass-based pasture of Brachiaria humidicola, an experiment to introduce legumes using seed-fixed macro-pellets was carried out in Llanos Orientales, Colombia. In August 1986, three kinds of promising legumes released by CIAT (Desmodium ovalifolium, Centrosema brasilianum and Arachis pintoi) were planted under the macro-pellet system in the grass-based pasture, following three types of land preparation, comprising minimum tillage with and without chemical grass-control, and nontillage with chemical grass-control. The planting density was 2,500 pellets/ha.
　The use of seed-fixed macro-pellets resulted in a satisfactory early-establishment of legumes regardless the differences of the method of seed-bed preparation, with an exception of A. pintoi introduced in the plot of nontillage with chemical grass-control. C. brasilianum vigorously grew after early-establishment, but was damaged by selective grazing. A. pintoi could not expand coverage due to its poor ability of competition with B. humidicola. The coverage of D. ovalifolium gradually expanded under grazing, and 16 months after planting, it reached over 60% of the whole area of both plots under the minimum tillage with and without chemical grass-control, while much lower coverage took place in the plot of nontillage with chemical grass-control. This result indicates that the combination of D. ovalifolium with minimum tillage would be the most adequate practice for introducing legumes in the pasture using macro-pellets.
　It was observed that the liveweight gains of cattle stocked in the B. humidicola and D. ovalifolium association were higher compared than those in the grass-based pasture. Remarkable liveweight gains (938 g/AU/day) were seen during the early wet season under the stocking rate of 1.8 AU/ha.
　Under the macro-pellet system adopted in this experiment, a very low application rate of chemical fertilizers (P: 2.0 kg/ha) was effective enough to secure the plant growth of legumes. In addition, the seed requirement of D. ovalifolium was 0.1 kg/ha. These requirements were only 5 to 10% of the conventional method. Such a low level of the input requirements has significant implications for reducing the costs in practice. It is therefore concluded that the macro-pellet method would play a key role in improving the productivity of grass-based pasture through the introduction of D. ovalifolium in the Llanos Orientales.