Historical data, physical findings, pulmonary function, arterial blood gases, and subjective degree of dyspnea rated on a modified Borg scale were correlated with eventual requirement of hospitalization in 83 episodes of acute asthma attacks of 70 adult patients. Among the pretreatment data, only pulse rate remained significant by a multivariate analysis to predict hospitalization. For patients who had apparently been successfully treated in the emergency room and discharged home, residual degree of subjective dyspnea was the only significant variable chosen by a linear discriminant function to predict the eventual need for hospitalization, with a sensitivity of 75% and a specificity of 78%. We concluded that careful clinical evaluation still remains the best available diagnostic tool in the care of acute asthma.