We compared the speed of action of the long acting beta-agonist salmeterol with that of salbutamol in order to assess whether reported in vitro differences are likely to have clinical significance. We used methacholine tests to produce a standardized level of bronchoconstriction and then observed the rate of recovery of FEV1 towards baseline after the administration by metered dose inhaler of salmeterol 50 micrograms or salbutamol 200 micrograms--doses which are considered to have similar bronchodilator potency. Twenty asthmatic subjects participated, and a double-blind, randomized, cross-over study design was followed. Salmeterol showed a significantly slower speed of action with median recovery to 90% and 95% of the baseline FEV1 (pre-methacholine) occurring after 9.6 and 19.4 min, respectively, compared with 4.8 and 8.3 min, respectively, for salbutamol (P less than 0.01). These observations are consistent with in vitro findings and suggest that salmeterol is likely to be less satisfactory than salbutamol as a 'rescue medication' for the treatment of acute episodes of bronchoconstriction.