A double-blind, randomized crossover study in 28 asthmatic patients assessed the relative therapeutic index for inhaled formoterol and salbutamol. Pre-drug administration FEV1 (mean 2.08 l) was 49-93% of predicted and reversibility 16-82% after inhalation of salbutamol. Patients inhaled single doses of formoterol (Oxis) (4.5,18 and 54 microg, delivered doses) via Turbuhaler, salbutamol (Ventolin) (200 and 1800 microg) via pressurized metered dose inhaler (pMDI) and placebo at intervals of 48 h or more. Individual maximum FEV1 and minimum S-K+ were calculated. Relative local (maximum FEV1) and systemic (minimum S-K+) dose potencies, and their ratio, the relative therapeutic index, were estimated using a non-linear mixed effect model. The drug effects were well tolerated and dose dependent. A log-linear approximation was used to describe the bronchodilatory effect, whereas a sigmoid approximation was more apt to describe the decrease in serum potassium concentration. A bivariate dose-response model based on these principles was fitted simultaneously to all data. The mean relative therapeutic index between formoterol 4.5-54 microg given via Turbuhaler and salbutamol 200-1800 microg given via pMDI was estimated to be 2.5 in favour of formoterol; this trend was not statistically significant.