A factorial experiment begun in 1980 included `Hamlin' and `Valencia' sweet-orange scions [Citrus sinensis (L.) Osb.], and Milam lemon (C. jambhiri Lush) and Rusk citrange [C. sinensis × Poncirus trifoliata (L.) Raf.] rootstocks, tree topping heights of 3.7 and 5.5 m, between-row spacings of 4.5 and 6.0 m, and in-row spacings of 2.5 and 4.5 m. The spacing combinations provided tree densities of 370, 494, 667, and 889 trees ha. Yield increased with increasing tree density during the early years of production. For tree ages 9 to 13 years, however, there was no consistent relationship between yield and tree density. Rusk citrange, a rootstock of moderate vigor, produced smaller trees and better yield, fruit quality, and economic returns than Milam lemon, a vigorous rootstock. After filling their allocated space, yield and fruit quality of trees on Milam rootstock declined with increasing tree density at the lower topping height. Cumulative economic returns at year 13 were not related to tree density.