The reliability of end-expiratory hydrogen (H2) breath tests were assessed and the significance of some important pitfalls were studied, using a compact, rapid H2-monitor with electrochemical cells. The H2 response was shown to be linear and stable. The reproducibility of the breath collection technique was determined in 20 patients following ingestion of lactulose. The increment between consecutive means of duplicate samples indicative of a significant rise of H2 concentration never exceeded 10 p.p.m., the mean coefficient of variation of the duplicate samples was below 5%. Fasting H2 levels were studied in 10 healthy adults during a 4-month period and they showed very marked inter- and intra-individual variability (16% above 40 p.p.m.). Initial peaks (early, short-lived H2 rises unrelated to carbohydrate malabsorption) were identified in 25% of the breath tests (in 4% above 20 p.p.m). It is concluded that the technique used for interval sampling of end-expiratory breath samples for H2 concentration gives reliable results. The biological significance of H2 concentration increments can only be evaluated if the limitations of the technical procedures and the individual ability to produce H2 is known.