An analysis of 1231 nests from 163 nest-sites recorded during a twenty-seven month long study (August 1990-NovemberI992) on the feeding ecology oflowland gorillas at the Bai Hokou Study Site, Dzanga-Sangha Dense Forest Reserve, Central African Republic is presented and compared with results from a concurrent nest-count census at the same site as well as results from an earlier census of the regional population and those from other gorilla populations. Nest counts are commonly used 1) to estimate population sizes and densities (Tutin & Fernandez, 1984; Carroll, 1986, 1988; Fay, 1988; Fay & Agnagna, 1992) and 2) to supplement direct observations and thus aid in understanding fluctuations in the size and composition of gorilla groups (Kuroda, 1992; Mitani et. ai., 1993). At the Bai Hokou Study Site, 44 % of gorilla nests recorded during the longitudinal study had no construction. At 15% of nest sites none of the nests were constructed. During a fine-grained census of the same study area, 30% of the nests identified were bare ground nests. These results contrast with reports from other sites of 0-10% bare ground nests (Tutin & Fernandez, 1984; Williamson 1988; Carroll, 1986; Fay, 1988; Schaller 1963). Nest construction type is probably related to the availability of suitable nesting materials at the nest site and at 8ai Hokou is shown to be influenced by micro-habitat, season and group size. These factors are also likely to affect interpopulation variability in nest construction. Bare ground nests and associated dung generally deteriorate to become unrecognizable after 4 days, while other nest construction types remain recognizable for more than 50 days (Tutin & Fernandez 1984, this study). Therefore, nest-counts will generally underestimate the numbers of nests at sites older than 4 days and may even fail to record entire nest sites. The accuracy of nest-count surveys may thus vary with location, habitats sampled and season.