1. Although bird censuses are imprecise, they have considerable value in a relative sense. 2. If the techniques are designed so as to minimize effects of variables that can be controlled, valid comparisons may be made among bird populations of two or more plots. 3. The spot-mapping method is recommended for greatest precision, and when a study is to be repeated over a period of years. Carefully standardized transects or point counts are most effective when many plots are being compared, and relative rather than absolute numbers will suffice. 4. Any method selected may be calibrated by a more thorough coverage of sample plots by the same or another method. 5. Demand for data on bird populations and species composition in specific habitats is increasing rapidly. Therefore, the results of census work should be published or at least made available through data banks or other means.