Prediction of the supply of glycogenic precursors to dairy cows and the site of degradation of wheat, maize and potato starch (PS) were determined in an in vivo experiment and the results were compared with data obtained from experiments involving in situ nylon bag and in vitro gas production techniques. In a Latin square design experiment four lactating dairy cows fitted with a rumen cannula and T-piece cannulae in the duodenum and terminal ileum, received either a low-starch control diet or diets in which sugar beet pulp in the concentrate mixture had been replaced by wheat, maize or PS. Starch from the different sources was almost completely degraded in the total gastrointestinal tract. For all starches, the rumen was the main site of degradation in vivo. No digestion of PS in the small intestine was observed. In situ results suggested that 14% of wheat starch (WS), 47% of maize starch and 34% of PS escaped rumen fermentation. According to the gas production technique WS ferments quickest and potato slowest. PS had a low degradability during the first 8 (gas production) to 11 (in situ) h. However, according to both in vitro and in vivo measurements rumen degradability of PS was high. The results suggest that in situ and in vitro techniques should be performed in animals that have adapted to starch source to provide a more accurate simulation of the in vivo situation.