The impact of large herbivores on the structure and composition of the woodlands in the Seronera Valley, Serengeti National Park, Tanzania, was reassessed in 1982 by updating an earlier survey by Croze (1974). Species composition, stand structure and the type and extent of browse damage were recorded in 22 open woodland stands dominated by Acacia tortilis, in association with Commiphora trothae, A. clavigera, A. senegal, or Albizia harveyi, and in eight riverine forest stands dominated by A. xanthophloea. Tree density ranged from 32 +/- 2 to 363 +/- 25 trees ha-1, and from 17 +/- 1 to 118 +/- 7 trees ha-1 among open woodland stands and riverine forest stands, respectively. Interstand differences in the frequency of damaged trees were a function of both species composition and population structure, but were independent of stand density. The relative frequency of occurrence correlated closely with the relative frequency of damage for each species, and a chi-2 distribution showed that only A. senegal was significantly preferred, while only C. trothae and A. clavigera were significantly avoided. The destruction of mature trees by elephants, and the intense browsing pressure on recruitment-sized trees by giraffes have significantly decreased the proportion of mature trees in the Seronera Valley over the past two decades. During the 10-year period prior to 1982, the frequency of mature A. tortilis and A. senegal trees greater than 5 m in height decreased by 70% and 89%, respectively, with an average annual mortality rate of 5% for mature A. tortilis trees. A comparison with Herlocker (1973) suggests that the density of A. tortilis in the valley in 1982 was approximately 41% less than that of similar communities in the region 10 years earlier. The density of A. xanthophloea in the Valley, and the frequency of large A. xanthophloea trees along the Seronera River declined by approximately 30% over this same period. Elephants within the Serengeti National Park have been poached to their lowest numbers in thirty years. Although the rate and direction of future woodland change in the Seronera Valley will be affected by unpredictable, recurrent elephant use, the frequency and intensity of fire, and the browsing pressure of giraffes are likely to be the most important factors. Accurate assessment of the impacts of these, and other factors on woodland regeneration has been hindered by the lack of comparable data describing these woodland communities. We believe that this data set used in conjunction with diligent monitoring of existing fire-management programmes will be essential to the understanding of long-term changes in vegetation structure in this ecosystem.