CONTEXT: Inhaled long-acting beta(2)-agonists improve asthma control when added to inhaled corticosteroid (ICS) therapy. OBJECTIVE: To determine whether ICS therapy can be reduced or eliminated in patients with persistent asthma after adding a long-acting beta(2)-agonist to their treatment regimen. DESIGN AND SETTING: A 24-week randomized, controlled, blinded, double-dummy, parallel-group trial conducted at 6 National Institutes of Health-sponsored, university-based ambulatory care centers from February 1997 through January 1999. PARTICIPANTS: One hundred seventy-five patients aged 12 through 65 years with persistent asthma that was suboptimally controlled during a 6-week run-in period of treatment with inhaled triamcinolone acetonide (400 microg twice per day). INTERVENTION: Patients continued triamcinolone therapy and were randomly assigned to receive add-on therapy with either placebo (placebo-minus group, n = 21) or salmeterol xinafoate, 42 microg twice per day (n = 154) for 2 weeks. The entire placebo-minus group was assigned and half of the salmeterol group (salmeterol-minus group) was randomly assigned to reduce by 50% (for 8 weeks) then eliminate (for 8 weeks) triamcinolone treatment. The other half of the salmeterol group (salmeterol-plus group) was randomly assigned to continue both salmeterol and triamcinolone for the remaining 16 weeks (active control group). MAIN OUTCOME MEASURE: Time to asthma treatment failure in patients receiving salmeterol. RESULTS: Treatment failure occurred in 8.3% (95% confidence interval [CI], 2%-15%) of the salmeterol-minus group 8 weeks after triamcinolone treatment was reduced compared with 2.8% (95% CI, 0%-7%) of the salmeterol-plus group during the same period. Treatment failure occurred in 46.3% (95% CI, 34%-59%) of the salmeterol-minus group 8 weeks after triamcinolone therapy was eliminated compared with 13.7% (95% CI, 5%-22%) of the salmeterol-plus group. The relative risk (95% CI) of treatment failure at the end of the triamcinolone elimination phase in the salmeterol-minus group was 4.3 (2.0-9.2) compared with the salmeterol-plus group (P<.001). CONCLUSIONS: Our results indicate that in patients with persistent asthma suboptimally controlled by triamcinolone therapy alone but whose asthma symptoms improve after addition of salmeterol, a substantial reduction (50%) in triamcinolone dose can occur without a significant loss of asthma control. However, total elimination of triamcinolone therapy results in a significant deterioration in asthma control and, therefore, cannot be recommended.