We measured gastric emptying of fat and water from a mixed meal in 6 healthy volunteers using a technique of duodenal perfusion that included phase-specific, nonabsorbable markers. Pairs of water-soluble markers (polyethylene glycol 4000 and phenolsulfonphthalein) and (equivalent) lipid-soluble ([14C]sucrose octaoleate and [3H]glycerol triether) markers were used. For each pair, one marker was used to label the corresponding component of the meal (water of fat), and the other was perfused into the duodenum as a recovery marker. The perfusion method was validated by comparing the amount of marker emptied, as estimated by marker dilution, with the amount obtained by complete aspiration of intestinal contents proximal to an occlusive balloon. Throughout the postprandial period, ratios of lipid marker to total fatty acids in the stomach remained constant, indicating a stable relationship of meal marker to total lipids. Water always emptied faster than did lipids, substantial amounts of which were still in the stomach after 6 h. There was close agreement between recovery of fat proximal to the balloon and calculations of lipid emptying from duodenal perfusion, supporting the validity of the marker dilution technique for the gastric emptying of lipids. This study demonstrates that when ingested as part of an ordinary mixed meal, water and fat have different patterns of emptying, water leaving the stomach first.