We used automated health insurance claims records of a New England insurer to assess the relation between salmeterol and severe nonfatal asthma. We identified 61,712 members who received a beta-agonist from January 1, 1993 to August 31, 1995, including 2, 708 recipients of salmeterol. Compared with recipients of other beta-agonists, future salmeterol recipients had higher rates of asthma hospitalization and dispensings of asthma medications during the year before they received salmeterol. We selected as a comparison group 3,825 recipients of sustained-release theophylline. We defined a baseline period as the year before the start of the follow-up period, and we characterized patients according to age, sex, calendar period, presence of baseline hospitalizations for asthma, presence of chronic obstructive pulmonary disease (COPD), and baseline dispensings of asthma medications. After adjusting for baseline factors, incidence rates of severe asthma in the salmeterol group were not elevated for emergency care (rate ratio estimate [RR] = 0.69, 95% confidence intervals [CI] = 0.42, 1.11), hospitalization (RR = 1.09, 95% CI = 0.60, 1.98), or intensive care unit (ICU) stays (RR = 0.81, 95% CI = 0.25, 2.62). We conclude that salmeterol was prescribed preferentially to high-risk patients and, after adjusting for baseline risk, salmeterol recipients did not have a greater risk than theophylline recipients of severe nonfatal asthma.