Ornithologists sampling breeding bird populations are subject to a number of biases in bird recognition and identification. Using Breeding Bird Survey data, these biases are examined qualitatively and quantitatively, and their effects on counts are evaluated. Differences in hearing ability and degree of expertise are the major observer biases considered. Other, more subtle influences are also discussed, including unfamiliar species, resolution, imagination, similar songs and attitude and condition of observers. In most cases, welltrained observers are comparable in ability and their differences contribute little beyond sampling error. However, just as hearing loss can affect results, so can an unprepared observer. These biases are important because they can reduce the credibility of any bird population sampling effort. Care is advised in choosing observers and in interpreting and using results when observers of variable competence are involved.