An approach for calculating bird densities from variable circular-plot counts is described. The approach differs from previous methods in that data from several surveys are pooled and detection distances are adjusted as if all distances were recorded by a single observer under a given set of field conditions. Adjustments for covariates that affect detection distances such as observer, weather, time of day, and vegetation type are made using coefficients calculated by multiple linear regression. The effective area surveyed under standard conditions is calculated from the pooled data set and then used to determine the effective area surveyed at each sampling station under the actual conditions when the station was sampled. The method was validated in two field studies where the density of birds could be determined by independent methods. Computer software for entering and analyzing data by this method is described.