Percent predicted peak expiratory flow (PEF) is used to determine the severity of asthma exacerbation and the appropriateness of discharge from the emergency department (ED). The 1995 Global Initiative for Asthma and 1997 National Asthma Education and Prevention Program guidelines recommend using a patient's "personal best" PEF, if available, as a better measurement than the predicted PEF obtained from population-based nomograms. We evaluated the accuracy of personal best PEF as provided by inner-city ED patients with acute asthma. One hundred four patients with acute asthma agreed to return to the ED for repeat PEF measurements on days 3, 7, 12, 21, and 24 after their initial ED visits for acute asthma. At the ED visit, only 29% (30/104) of patients were able to report a personal best PEF. Over the 24 days of follow-up, 45% (10/22) of these patients had a measured PEF greater than their reported personal best. If a predicted PEF of at least 70% was used as the criterion for ED discharge, as several asthma guidelines recommend, then using patients' reported personal best PEF would have led to inappropriate ED discharge for some patients.