Data collected in a 40-ha reedbed of southern France were used to compare the efficiency and limitation of mist-net and point-count techniques in estimating the composition and structure of a bird assemblage dominated by the Bearded Tit (Panurus biarmicus), the Moustached Warbler (Acrocephalus melanopogon), and the Reed Warbler (Acrocephalus scirpaceus). Null model analyses were used to determine the effect of spatial variability on estimates of species richness and relative abundance with the two sampling techniques. A 50-m net line operated during 5 h or two 50-m radius point counts of 10 min conducted 6 wk apart provided a similar estimation of species composition and relative abundance. While a sampling effort of 10 net lines or 13 point counts would permit the detection of a 25% difference in the relative abundance of most species (whether over time or among sites), the analyses on community structure suggest that 26 net lines or 13 point counts are necessary to sample adequately the structure of the bird assemblage (proportion of individuals from each species) in the 40-ha reedbed. This minimum sampling effort, however, is strongly influenced by the variance in number of individuals sampled, which differed among bird species. For instance, removal of highly mobile species such as the Bearded Tit, reduced by half the minimum sampling effort. To provide a reliable estimate of species richness and relative abundance, point counts must be based on both visual and auditory cues (30% of the birds sampled were silent), and be carried out at different periods to cover the early breeding season of any abundant species. While point counts are less demanding in the field, manipulation of netted birds permits the documentation of various aspects of their biology, which compensated for the increased time and effort needed.