We estimated statewide, regional, and national trends in counts of Bald Eagles (Haliaeetus leucocephalus) along selected routes in the contiguous United States during midwinter, 1986-2000. Each January, several hundred observers collected data as part of a survey initiated by the National Wildlife Federation in 1979. To analyze these data, we used only those routes surveyed consistently in at least four years and on which at least four eagles were counted in a single year. We included surveys conducted during fog or precipitaion after determining that changes in weather conditions probably did not affect trend estimates. Our final analysis, using a hierarchical mixed model, was based on 101,777 eagle sightings during 5,180 surveys of 563 routes in 42 states. In the model, fixed effects were year, region, and route-length category; the random effect was the route itself. Model-based estimates of Bald Eagle counts throughout the U.S. increased 1.9% yr^-1, but trend estimates varied by region. Estimated trends were statistically significant, and positive, in the northeastern U.S. (6.1% yr^-1), but were not significant in other regions. The proportion of increasing counts was higher north of 40° N and east of 100° W. Trends in numbers of adults and immatures showed similar geographic patterns, but counts of adults increased at a higher rate. Overall, trends were more similar to those identified by the Christmas Bird Count than the Breeding Bird Survey. In spite of limitations, the survey is a cost-effective way to monitor wintering eagles in the lower 48 states. We discuss estimated trends in the context of increased urbanization, changed weather, and recovery from pesticide pollution.