The giant tortoise ( Geochelone gigantea (Schweigger)) population of Aldabra has varied greatly in numbers since the beginning of the present century. Recent estimates have shown that the population is composed of 150 466 ±s.e. 16 441 animals. Of this total, 60% are located in an area of 33.6 km 2 at the eastern end of Grande Terre. Animals composing the small population of 2000 tortoises on Malabar grow continuously, while those on Grande Terre only grow seasonally and are much smaller than their less numerous counterparts on Malabar. It is suggested that shade factors limit the time available for feeding in Grande Terre. Records of movement show that while some animals do move large distances 53% of the population are not relocated more than 500 m from their initial marking point. A mean annual rainfall of 941 mm would be expected to yield 1887 g m -2 a -1 (dry mass) of primary production, with a range of from 2337 to 4037 g m -2 a -1 . Tortoises of average mass (20-30 kg) consume 79 kg a -1 . Estimates of total consumption for areas with differing tortoise densities suggest that they would consume 11.3% at the eastern end of Grande Terre and 0.7% in the Pemphis scrub. Defecation records suggest that the gross assimilation efficiency of giant tortoises is about 50 %. The mean mass of tortoises on Aldabra are 21.7 kg on Grande Terre, 49.9 kg on Malabar and 51.3 kg on Picard. Standing crop biomasses derived from these weights are 35 387 kg km -2 on Grande Terre, 35 084 kg km -2 on Malabar and 25 342 kg km -2 on Picard. These biomass data are significantly higher than those achieved by large herbivores on mainland African wildlife ecosystems. By using data available on biomass mortality, P/B ratios (turnover times) of 0.042 for Grande Terre and 0.034 for Malabar are obtained. Annual production calculated from these ratios are 1486 kg km -2 a -1 for Grande Terre and 1193 kg km -2 a -1 for Malabar. The production efficiency of the giant tortoise population is about 2.1% which is in close agreement with figures obtained for other long-lived poikilotherms. Potential production for Grande Terre predicted from the mean rainfall only differs by 2.6% from that estimated. The eastern end of Grande Terre, however, exceeds this predicted figure by 61% and it is suggested that this is due to increased primary production induced by water available from the freshwater lens raised by spring tides. This phenomenon is similar to mainland African wildlife ecosystems fed by abundant ground water.