To evaluate the accuracy of allergy histories obtained by interns, residents, and ward attending physicians, we compared physician-charted allergy histories with our own concurrently performed histories over a 2-month span for all general medicine and surgery patients admitted to a 615-bed suburban teaching hospital. We found 100% agreement between investigator and charted histories for all 318 (88%) of 363 charts identifying no known allergy. We disagreed with 23 (50%) of 46 allergies documented in the charts of the remaining 45 patients. Concurrent review classified 18 (39%) of these 46 charted allergies as clinically unimportant, and 5 charts (11%) were deemed underclassified. A description of the allergy by the primary physician was documented in only 7 charts (16%). Inaccurate allergy histories are frequently documented in the medical record. Simple adjustments in history taking and supervision may have important implications for patient care and training.