Selected secondary diagnoses (e.g. pulmonary embolism) may provide an efficient and inexpensive source of data for quality assurance (QA) monitoring if their absence at admission were known. In June 1990 we modified our hospital abstracting methods to classify each diagnosis into categories: (1) present on admission, (2) acquired during hospitalization, or (3) uncertain. Our experience has confirmed the identification and elimination from QA reports of the majority of pre-existing secondary diagnoses. Examples of secondary diagnosis codes acquired or uncertain were acute myocardial infarction 48%, pneumonias 25%, pulmonary emoboli 54% and cerebral vascular accident/hemorrhage 35%. Abstracting time has increased less than 2 min per discharge. A reabstraction study showed 87% agreement (kappa = 0.733, p less than 0.001) between initial collection and blinded reabstraction. The separation of secondary diagnoses into preexisting or acquired can: (1) be reliably undertaken by discharge abstracters; (2) be efficient in adding minimal time; and (3) enhance the validity and usefulness of data and increase physician acceptance.