Coffee plantations in Brazil are usually made with four seeddlings per hill. To study the problem of spacing the seedlings in the hill, an experiment is being carried out at the Estação Experimental Central of the Instituto Agronômico in Campinas in the "terra-roxa-misturada" type of soil. Four square spaciugs between the seedlings were compared: 10, 20, 30 and ±0 cm. Yield data and other observations reported in this paper refer to the first four harvests. The following results have been so far recorded: 1. Measurements of the seedlings made at three different times indicated that there was a significant statistical difference between plants in the same hill, and that seedlings set 10 or 20 cm apart were higher than those planted at 30 or 40 cm. 2. The statistical analysis of yield data showed a significant linear increase related to the increase in spacing between seedlings in the hill. 3. Variations in the spacing between seedlings in the hill did not affect the size of the coffee seeds.