Before endoscopy a double-blind, randomized, controlled, single-subject trial comparing the symptomatic effect of 1-day treatment periods with cimetidine and placebo was conducted in patients with dyspepsia. Results from 339 patients were analysed. The trial lasted 12 days and consisted of 6 treatment days with 400 mg cimetidine three times daily and 6 days with placebo three times daily. The order of the treatments was randomized within six pairs, and a randomization test based on daily measures of global symptoms provided individual p values. Aggregation of the measures from all subjects showed that cimetidine alleviated the symptoms significantly better than placebo in peptic ulcer disease (PUD) (p less than 0.0001), oesophagitis (p less than 0.001), and non-ulcer dyspepsia (NUD) (p less than 0.0001). Twenty-seven per cent of the patients with PUD, 26% of those with oesophagitis, and 12% of the patients with NUD obtained individual p values of less than 0.10 and were defined as responders. The best predictors of the response to cimetidine in NUD were age above 40 years, heartburn or acid regurgitations being the worst symptom, and night pains relieved by food, milk, or antacids. In conclusion, the applied single-subject trial confirmed the overall symptomatic effect of cimetidine in dyspepsia and identified individual responders among patients with NUD with a clinically reasonable profile. The low proportion of responders among patients with PUD or oesophagitis suggests that the model has a low sensitivity for identification of individual responders and that the single-subject trial design in dyspepsia needs further refinement.