One of the major purposes of bird population studies is to document changes in population size over a period of years. The traditional method used in Europe and North America to detect population change is to calculate annual ratios. However, this method can produce spurious results when ratios are accumulated over many years. Consequently, new methods of computing trends are needed. Several new methods of estimating population trends are developed and illustrated with data from the North American Breeding Bird Survey (BBS). Each method is compared in terms of its assumptions, biases, and limitations. On the basis of these comparisons we recommend one method that we feel most accurately detects true population trends. Both the biological and statistical justifications for the model selection are presented. Trends estimated with this model are then presented for two species.