Abstract We established an experiment in 1993 to test the effect of various treatments on growth of a native timber tree, Terminalia amazonia, on eroded pasture in southern Costa Rica. The 1993 treatments included a control in which T. amazonia was planted alone, and an interplanted treatment with the legume tree Inga edulis. Measurements at 4, 8, and 11 years showed that T. amazonia grew significantly better when mixed with Inga. In 2004 (11 years) we tested whether improved nitrogen nutrition was the mechanism for the effect. We analysed standing crops of vegetation and their nitrogen content in controls, the interplanted treatment, and nearby grazed pasture. Foliar N in Inga was more than double that in other foliage, supporting our assumption that the legume could be a source of N. However, by 2004 soil N was only slightly higher in the interplanted treatment than in controls, suggesting that N might be incorporated rapidly by the other vegetation. In fact, when the interplanted treatment was compared to controls, above-ground biomass and foliar N of T. amazonia was higher. The total nitrogen content in crowns of T. amazonia (biomass × % foliar N) was higher in the interplanted treatment than in controls. Finally, basal diameter and height correlated with foliar N in T. amazonia. Our data were consistent with the hypothesis that improved nitrogen nutrition accounts for improved growth of T. amazonia when interplanted with I. edulis. Total volume production of the crop tree T. amazonia was too low for forestry goals. However, interplanting the two tree species in most cases provided promising beginnings of complexly structured secondary forest for wildlife habitat.