1 - The present paper deals with the first results of an extensive study on the relations between stock and scion in Citrus, in progress at the Experiment Station of Limeira, of the Instituto Agronômico of Campinas, São Paulo, Brazil, since 1933. 2 - All the scions used belong to three commercial varieties : Baianinha (orange, derived from "Washington Navel"), Pera (orange, similar to Valencia) and Pomelo (grape-fruit "Marsh-seedless") were collected in each case from one individual of each type. 3 - The following 12 stock were studied : Laranja azeda (sour orange) Laranja agro-doce (bitter sweet orange) Laranja caipira (sweet orange) Laranja lima (sweet orange) Tangerina cravo (tangerine) Limão cravo (Rangpur lime) Limão rugoso (rough lemon) Limão ponderosa (ponderosa lemon) Lima da Pérsia (sweet lime) Pomelo (grape-fruit Triumph) Cidra (citron) Trifoliata (Poncirus trifoliata). All plants obtained came from seeds of one tree of each type, from unprotected flowers however. 4 - Data referring to the number of seeds per fruit, germination, polyembriony and classification according to size are given (quadro I). The mean height of the plants obtained after 6 months and 2 years nursery are to be found in quadro II. 5 - Twelve hundred individuals of each root stock were selected according to their uniform development, in accordance with Webber (25) and on 40 of each buds of the three scions were grafted. As measure of the development the circumference of the stock at a height of 20 cm and that of the scion at a height of 40 cm were used (quadros IV to IX). 6 - An analysis of variance "between" and "within" root stocks (quadro VI) showed in all cases a significant high variation "between" stock. In all cases, the scions were thinner than the stocks (quadro IX), the indices scion-stock varying from ca. 0,5 to 0,8. The respective means are to be found in quadro IV and V. 7 - There is also a consistent and significant difference between scions as shown by the values in quadro VII. 8 - A comparison between quadro VIII (not budded root stock) and quadro IV (budded) shows that generally the former grow better. They are however some exception to this rule, especially in "Tangerina cravo". 9 - All the observation reported so far were made in the nursery, but the detailed lay-out of the experiment in the orchard is also described. Lots of 4 or 9 plants of each combination were arranged in four complete random replications. The trees supported the planting out very well, only Citron replantings being necessary. The trees are now fully matured and produced already during 3 years. The results of the first five years of the experiment with trees in the orchard, including the analysis of their yields, will be given in a future publication.