Background: Since 1980, 21 persons have died from rabies acquired in the United States. Eighteen (84%) of the deaths were caused by rabies virus associated with bats. However, information concerning human encounters with bats and the risk ibi rabies is limited. Methods: We analyzed state laboratory records of all bats submitted foi rabies diagnosis during 1977-1996, and studied all case reports of verified encounters with bats to characterize the circumstances of encounters with humans. An encounter was defined as any time that a bat was in close proximity to a person. Results: From the laboratory records, the overall prevalence of rabies in bats was 15% (681/4470). Bats that bit humans were more likely to have had rabies than those that did not bite humans (30% [69/233] versus 14% [612/4237]; prevalence ratio=2-l ; 95% confidence interval=1.7-2.5). From 239 case reports, 52% of the encounters took place around the home, and 83% occurred during June-Septembei The most frequently reported circumstances were a bat landing on an awake person (19%), a person picking up a grounded bat (18%), awakening to find a bal in the room ( 15%), and trying to remove an indoor bat ( 10%) Conclusions: There is a substantial threat of rabies from bats, and many human encounters with bats are preventable. The results support the Advisory Committee on Immunization Practices draft recommendation dated June 1996, which stated that persons with even suspected exposure to bats should receive post-exposure prophylaxis unless prompt diagnosis excludes rabies. Improved public awareness of the risk for rabies from bats could limit preventable encounters and thereby reduce the need for prophylaxis.