Dispersal is a critical link between organismal and population biology, yet, because of their mobility, our understanding of the causes and consequences of long-distance dispersal by birds remains poorly known. Methods used to study dispersal include (1) marking and recapturing individuals in a limited study area to estimate survival and dispersal rates, and (2) relying on volunteers to mark and recapture individuals over larger areas. We compared these two methods for measuring dispersal distances of Tree Swallows (Tachycineta bicolor) using recapture data from a limited-area study in New York State (the Swallow Dispersal Study, SDS) and the recapture dataset from the U.S. Bird Banding Laboratory (BBL). Analysis of BBL records revealed a difference in the dispersal distance distributions (DDD) for data reported before and after 1967. In the earlier data, 84% of the 238 records were for birds within 13.6 km of their first banding location, whereas only 22% of the 799 records in the more recent data were reported in this closest distance belt. These differences are almost certainly due to changes in reporting protocols instituted by the BBL in the mid-1960s. We corrected for recapture effort in the SDS, and, using this corrected SDS data for the proportion of birds returning in the closest distance belt and the recent BBL for the proportions of more distant movements, we created what we think is the best composite DDD for Tree Swallows. Even though dispersal distances up to 2367 km have been reported, the composite DDD indicates that fewer than 3% of birds disperse more than 100 km and that 85% disperse less than 15 km between years. Thus, our results suggest that the dispersal behavior of most individuals can be examined effectively at more local spatial scales. Studies of dispersal and mortality would be facilitated if all recaptures of banded birds were reported with accurate spatial coordinates to the Bird Banding Laboratory.