The symptomatic effect of cimetidine was examined in 27 patients with non-ulcer dyspepsia (NUD) by means of a multi-cross-over model (MCO model) for testing the symptomatic effect of drugs in individual patients. None of the patients showed an ulcer at the time, but 20 patients had evidence of previous peptic ulcer disease. The variant of the MCO model used included six treatment periods and three regular interchanges between cimetidine and placebo. Treatment periods lasted 2 or 4 days. The individual results were evaluated on the basis of the number of times (X score) cimetidine was associated with less symptoms than the preceding or following placebo. In general, cimetidine was associated with significantly (p less than 0.02) less symptoms than placebo. The X-score distribution was therefore skew in favour of high scores. Five patients showed the maximal X score of 5. The chance of getting an X score of 5 when cimetidine is not better than placebo is about 9%. Accordingly, the risk of being wrong when defining these five patients as cimetidine responders is 9%. The present study confirms that the MCO model may identify individual cimetidine responders among patients with NUD.