The effect of gastrectomy and duodenal reflux on gastric carcinogenesis was studied because gastrectomized patients may be considered at "high risk" for the development of gastric stump cancer. Wistar rats received N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) (83 mg/liter) ad libitum in the drinking water for either four, eight, or twelve weeks. A control group received tap water. After MNNG administration animals were antrectomized. Antrectomy was not performed in a control group. Bowel continuity was restored either with a Billroth II (BIL) or with a ROUX en Y (ROUX) procedure. Duodenogastric reflux is possible after the BIL but not after the ROUX procedure. Eight months after the beginning of the experiment the stomachs of the animals were studied. In both operated and unoperated animals, the number of cancers observed was significantly related to the duration of MNNG administration. Animals receiving MNNG plus the BIL procedure had a significantly higher number of anastomotic cancers than the ROUX animals, indicating that duodenogastric reflux played a promotional role in gastric carcinogenesis. Three BIL gastrectomized rats not receiving the carcinogen had an adenocarcinoma on the anastomotic line further emphasizing the risk attached to the duodeno-gastric reflux.