This study aimed to examine in vivo starch digestion kinetics and to unravel the mechanisms of starch hydrolysing enzymes. Ninety pigs (23 (sd 2·1) kg body weight) were assigned to one of nine treatments in a 3×3 factorial arrangement, with starch source (barley, maize, high-amylose (HA) maize) and form (isolated, within cereal matrix, extruded) as factors. We determined starch digestion coefficients (DC), starch breakdown products and digesta retention times in four small-intestinal segments (SI1–4). Starch digestion in SI2 of pigs fed barley and maize, exceeded starch digestion of pigs fed HA maize by 0·20–0·33 DC units (P<0·01). In SI3–4, barley starch were completely digested, whereas the cereal matrix of maize hampered digestion and generated 16 % resistant starch in the small intestine (P<0·001). Extrusion increased the DC of maize and HA maize starch throughout the small intestine but not that of barley (P<0·05). Up to 25 % of starch residuals in the proximal small intestine of pigs was present as glucose and soluble α(1–4) maltodextrins. The high abundance of glucose, maltose and maltotriose in the proximal small intestine indicates activity of brush-border enzymes in the intestinal lumen, which is exceeded by α-amylase activity. Furthermore, we found that in vivo starch digestion exceeded our in vitro predictions for rapidly digested starch, which indicates that the role of the stomach on starch digestion is currently underestimated. Consequently, in vivo glucose release of slowly digestible starch is less gradual than expected, which challenges the prediction quality of the in vitro assay.