The US National Institutes of Health (NIH) guidelines for treating asthma classify patients into four severity levels (severe persistent, moderate persistent, mild persistent, and mild intermittent) based largely on clinically assessed symptoms and physical measurements such as forced expiratory volume. The guidelines also recommend specific regimens of various asthma medications for each severity level. OBJECTIVE: To use the medication regimens recommended by NIH to classify asthma severity based on patterns of drug use in prescription claims. METHODS: Using claims data from a large health insurer in the Northeast US, study patients were selected in part on the basis of having used zafirlukast continuously at some point during the 30 months covered by the database. Of the 780 study patients, 76.8% could be classified by a computer program into one of the four NIH severity categories based on their patterns of asthma drug use as specified in the NIH guidelines. The remaining 23.2% were classified through review of individual claim histories by a nurse abstractor. RESULTS: 60.9% of study patients were classified as having mild persistent asthma, 22.6% were moderate persistent, 10.4% were severe persistent, and 6.2% were mild intermittent. The nurse abstractor also performed a blind review of the claim histories for a 15% sample of patients who had been programmatically classified. The two classifications disagreed in only six cases (i.e., 93.3% agreement, kappa = 0.885, p < .001). CONCLUSION: Asthma patients can be classified by severity level based on drug use patterns in prescription claims with an acceptable level of validity based on clinical review. This methodology could facilitate efficient monitoring of asthma severity levels within patient populations.