Summary1. Aerial and ground counts in the Queen Elizabeth National Park from 1963 to 1969 indicate that the number of elephants has more than doubled. The increase appears to be the result of immigration rather than reproduction.2. Elephant food habits were studied for 10 months in two areas of the Park. In the short‐grass/thicket area, browse intake rose markedly whenever the rainfall fell below 50 mm/month. In the tall‐grass area without thicket, herbs were eaten frequently during the rains and Cymbopogon grass bases in the dry season.3. Food availability had an important influence on the diet of elephants. Sometimes, however, elephant were very selective, e.g. for Panicum maximum, Azima tetracantha, Securinega virosa, and Tribulus terrestris. Certain grass inflorescences were selected during the rains, and bases in the dry season.4. Most browse was consumed in the thicket area during the dry season and most grass was eaten in the tall‐grass area of the Park.5. Stomach and faecal samples gave a similar result and indicated that the diet in the tall‐grass area was comparable to that of elephant in Murchison Falls National Park.6. Analyses of important plant species showed that, in general, tall‐grass had a lower crude protein content than short‐grass, herbs and browse. Browse leaves had a relatively high crude‐protein content in the dry season. Differences were less marked in the rains.7. Ether extract, or fat content, was high in browse leaves and in Cymbopogon, in particular its bases. These may have been selected in the dry season for this reason.8. Grass and browse stems were very fibrous and this may have contributed to their being avoided.9. High carbohydrate levels probably enhanced the palatability of grass bases and browse leaves.10. Grasses were very siliceous, in particular their bases when contaminated with soil. Elephant took considerable trouble to remove soil. Browse leaves had little silica but were rich in other minerals, and this may have contributed to their greater palatability.11. Studies of woodland dynamics show a decline in large trees which corresponds to the increase in elephants. In some cases trees have been almost eliminated. Damage to trees indicates that a most important factor in their decline is the elephant. There is some evidence that the trees are eliminated selectively.12. In the tall‐grass area there has been an increase of 46.3% in elephant numbers, but a decline of 80.3 % in buffalo, over the past 15 years. An overall decline in combined biomass of 36.9 % is calculated. Estimates of energy utilization of the annual vegetation production by elephant and buffalo indicates a decline from 6.1 % to 3.9% over this period.