Starch, not fibre, is probably the major substrate for fermentation in the human colon. However, quantitating the amount of starch that resists pancreatic amylase and thus escapes digestion in the small bowel is difficult. A number of techniques have been employed in man and are reviewed here, including direct intubation of the ileum, the ileostomy model, and breath studies. The results of a series of studies of the digestion of starch from potato and banana are reported. When fed to ileostomy patients, 3% of hot potato starch and 12% of cold potato starch were resistant to digestion, as was 75% of banana starch. In feeding experiments with healthy volunteers none of the starch was recoverable in faces, indicating its complete fermentation in the colon. Breath H2 measurements after test meals of these starches indicated that only 2-5% of potato starch and 7-12% of banana starch was fermented. A single blood acetate measurement timed to coincide with peak breath H2 was not useful. However, a number of problems with breath H2 studies are discussed, and it is suggested that either ileal intubation or the ileostomy model are the most reliable techniques presently available, with serial blood acetate determinations also potentially valuable. Overall on Western diets, approximately 10% of all starch is probably resistant starch.