Many asthma patients remain symptomatic despite maintenance therapy with inhaled corticosteroids (ICS) and salbutamol as rescue medication. In the present study the relative efficacy and preference for as-needed formoterol compared with salbutamol was examined. In total, 211 patients with a mean age of 45 yrs (mean forced expiratory volume in one second (FEV1) 77% predicted normal), using ICS, were randomised to 3 weeks' double-blind treatment with as-needed formoterol 4.5 microg Turbuhaler and with as-needed salbutamol 100 mug Turbuhaler in a cross-over fashion. Overall, lung function and symptom control were better with as-needed formoterol than with as-needed salbutamol. During as-needed formoterol treatment daytime and night-time symptom scores were lower, peak expiratory flow and FEV1 were higher and patients experienced fewer disturbed nights (34%) compared with as-needed salbutamol. Patients preferred the formoterol treatment to salbutamol. Of the 162 patients expressing a preference, formoterol was preferred by 68% (95% confidence interval: 60-75). Subjective assessment of effectiveness also favoured formoterol, which was perceived as slightly faster acting than salbutamol. In conclusion, as-needed formoterol improved symptoms and lung function compared with salbutamol and was perceived as more effective and at least as fast acting for symptom relief.