The technique of sampling fish in coves by the application of rotenone ("covesampling") was evaluated in Barkley Lake, Kentucky, 26-28 September 1978. In 85 ha Crooked Creek Bay (which was subdivided into 24 coves and other subareas), 776 kg/ ha of fish were recovered; the total adjusted for non-recovery of marked fish was 865 kg/ ha. The numbers and biomass of fish in various coves within the Bay did not closely approximate the standing crop in the total area. Most major species of fish were either overrepresented or underrepresented in the cove samples. Larger coves (mean area, 4.9 ha) produced a more representative sample of the 85-ha Bay with respect to species composition, abundance, standing crop and length distribution. Results of the Barkley Lake study and those of a similar study in Douglas Lake, Tennessee, in 1965 indicated that many adjustment factors are required to equate small-cove standing crop estimates to those of a large bay or an entire reservoir.