Variation in species richness at the landscape scale is an important consideration in conservation planning and natural resource management. To assess the ability of rapid inventory techniques to estimate local species richness, three collectors sampled the spider fauna of a "wilderness" cove forest in the southern Appalachians for 133 person-hours during September and early October 1991 using four methods: aerial hand collecting, ground hand collecting, beating, and leaf litter extraction. Eighty-nine species in 64 genera and 19 families were found. To these data we applied various statistical techniques (lognormal, Poisson lognormal, Chao 1, Chao 2, jackknife, and species accumulation curve) to estimate the number of species present as adults at this site. Estimates clustered between roughly 100-130 species with an outlier (Poisson lognormal) at 182 species. We compare these estimates to those from Bolivian tropical forest sites sampled in much the same way but less intensively. We discuss the biases and errors such estimates may entail and their utility for inventory design. We also assess the effects of method, time of day and collector on the number of adults, number of species and taxonomic composition of the samples and discuss the nature and importance of such effects. Method, collector and method-time of day interaction significantly affected the numbers of adults and species per sample; and each of the four methods collected clearly different sets of species. Finally, we present recommendations to guide future research on the estimation of spider species richness.