Study objectives were to evaluate the 1-hour decision point for discharge or admission for acute asthma; to compare this decision point to the admission recommendations of the Expert Panel Report 2 (EPR-2) guidelines; to develop a model for predicting need for admission in acute asthma. The design used was a prospective preinterventional and postinterventional comparison. The setting was a university hospital emergency department. Participants included 50 patients seeking care for acute asthma. Patients received standard therapy and were randomized to receive albuterol by nebulizer or metered-dose inhaler with spacer every 20 minutes up to 2 hours. Symptoms, physical examination, spirometry, pulsus paradoxus, medication use, and outcome were evaluated. Based on clinical judgment, the attending physician decided to admit or discharge after 1 hour of therapy. Outcome was compared to the EPR-2 guidelines. Post hoc statistical analyses examined predictors of the need for admission from which a prediction model was developed. Maximal accuracy of the admit versus discharge decision occurred at 1 hour of therapy. Using FEV(1) alone as an outcome predictor yielded suboptimal performance. FEV(1) at 1 hour plus ability to lie flat without dyspnea were the best indicators of response and outcome. A model predictive of the need for admission was developed. It performed better (P =.0054) than the admission algorithm of the EPR-2 guidelines. The decision to admit or discharge acute asthmatics from the ED can be made at 1 hour of therapy. No absolute value of peak flow or FEV(1) reliably predicts need for hospital admission. The EPR-2 guideline thresholds for admission are barely adequate as outcome predictors. A clinical model is proposed that may allow more accurate outcome prediction.