Recent studies have suggested that regular use of inhaled beta 2 agonists cause loss of asthma control as measured by worsening peak-flow rates, increased asthma symptoms, and more frequent need for supplementary bronchodilators. However, the magnitude of this effect and the reliability of investigator-originated definitions of control is unknown. We studied 341 people with asthma in a four-week, randomised, crossover trial of regular salbutamol (2 puffs--200 micrograms--four times daily) for two weeks and as needed for two weeks. There were no significant differences in morning and evening peak-flow rates between treatments but asthma symptoms and supplementary bronchodilator use were significantly less frequent when salbutamol was given regularly. Asthma episodes occurred 1.39 (1.52) times per day during regular treatment and 2.44 (1.75) times per day during as-needed treatment (p < 0.0001) and 0.50 (0.56) vs 0.65 (0.66) times per night (p < 0.0001). Daytime use of supplementary salbutamol was 1.14 (1.40) vs 2.35 (1.71) puffs per day, (p < 0.0001); night-time use was 0.45 (0.55) vs 0.64 (0.66) puffs per night (p < 0.0001). When control endpoints were compared between treatment periods for each individual by two blinded investigators and control judged by six different sets of criteria, in 70 asthmatics there was no difference in symptom control between periods but in the remainder, control was achieved more often by regular than by as-needed salbutamol (166 vs 69, p < 0.0001). In asthma of moderate severity, regularly administered salbutamol does not produce lower peak flow rates than as-needed salbutamol and is associated with less frequent asthma symptoms.