OBJECTIVES: To assess the effect of montelukast versus salmeterol added to inhaled fluticasone propionate on asthma exacerbation in patients whose symptoms are inadequately controlled with fluticasone alone. Design and setting A 52 week, two period, double blind, multicentre trial during which patients whose symptoms remained uncontrolled by inhaled corticosteroids were randomised to add montelukast or salmeterol. PARTICIPANTS: Patients (15-72 years; n = 1490) had a clinical history of chronic asthma for > or = 1 year, a baseline forced expiratory volume in one second (FEV1) value 50-90% predicted, and a beta agonist improvement of > or = 12% in FEV1. MAIN OUTCOME MEASURES: The primary end point was the percentage of patients with at least one asthma exacerbation. RESULTS: 20.1% of the patients in the group receiving montelukast and fluticasone had an asthma exacerbation compared with 19.1% in the group receiving salmeterol and fluticasone; the difference was 1% (95% confidence interval -3.1% to 5.0%). With a risk ratio (montelukast-fluticasone/salmeterol-fluticasone) of 1.05 (0.86 to 1.29), treatment with montelukast and fluticasone was shown to be non-inferior to treatment with salmeterol and fluticasone. Salmeterol and fluticasone significantly increased FEV1 before a beta agonist was used and morning peak expiratory flow compared with montelukast and fluticasone (P < or = 0.001), whereas FEV1 after a beta agonist was used and improvements in asthma specific quality of life and nocturnal awakenings were similar between the groups. Montelukast and fluticasone significantly (P = 0.011) reduced peripheral blood eosinophil counts compared with salmeterol and fluticasone. Both treatments were generally well tolerated. CONCLUSION: The addition of montelukast in patients whose symptoms remain uncontrolled by inhaled fluticasone could provide equivalent clinical control to salmeterol.