The aim was to determine diet selection of goats grazing/browsing in Zululand Thornveld. An experiment was done in spring (November 2007) and autumn (March/April 2008) at Owen Sitole College of Agriculture, KwaZulu-Natal. Diet selection was estimated using direct observation of animals and an indirect plant-based method. The browse species observed in the plant-based methods were: Gymnosporia senegalensis (Gs), Acacia nilotica (An), Acacia karroo (Ak), Scutia myrtina (Sm) and Dichrostachys cinerea (Dc). Selection was determined by observing at regular intervals 40 marked branches on 10 plants of each browse species. Direct observation of diet selection was done by observing two goats in a base herd of 24 castrated males when they were allowed to forage together in a paddock of 1 ha. Observations were made every minute for three hours in the morning and 1.5 hours in the afternoon. The relative abundance of the browse species in the veld was determined and used to estimate the selection index (utilization relative to availability). In both seasons, the three most selected species according to the plant-based observations were Sm, Ak and Dc. Consistently, An experienced moderate defoliation and Gs the least. However, on the basis of the selection index the species followed the order: An>Dc>Ak>Sm>Gs in spring and An>Sm>Dc>Ak>Gs in autumn. The selection index was negatively correlated to all fibre attributes although the correlation attained significance only for NDF, ADF and lignin but not for tannins and cellulose. Both methods did not rank species in the same order, thus suggesting the need for further methodological refinement.