We used a three-year point-count data set from Hoosier National Forest, Indiana, to evaluate alternative point-count sampling strategies for detecting songbird declines. Repeated-measures ANOVA indicated that mean abundance estimates increased with increasing count radius (P < 0.0001, each species), and coefficients of variation (CVs) decreased. Mean abundance estimates increased with longer count duration (6-, 8-, and 10-min) for only two of 13 species, and CVs did not necessarily decrease. The power to detect a 5% annual population trend increased with more survey points, more visits per point, and more years of surveys. Managers can use observer time most efficiently by employing counts of short duration. Counts using a larger radius will reduce CVs and therefore provide potentially better annual estimates of abundance and power to detect changes over a period of years. The design of the study and nature of the variability in bird abundance will determine whether increasing the number of points, or the number of visits per point, will have greater effect on power to detect a population trend.