What should we expect from pilot studies, done in small series of patients? In the literature there are many examples of small studies with very promising results, that in subsequent larger or better controlled studies proved to be much less promising, or even disastrous. In some instances the initial favourable outcome was due to selection bias. In others the use of nonvalidated methods of measuring outcome made the reproducibility of promising observations problematic. However, we have to start somewhere. In ths issue The et al. report favourable results of granisetron treatment in four out of five patients with chronic fatigue syndrome. A prospective, randomised, placebo-controlled, double-blind clinical trial with granisetron in patients with chronic fatigue syndrome is now ongoing.