A dynamic model of the human upper gastrointestinal (GI) tract was designed to better simulate conditions of ingestion and digestion, by including a food matrix as part of the model design. The dynamic model consisted of two reactors maintained at 37 °C, one simulating stomach conditions and the other simulating duodenum conditions. The model was tested by comparing survival of bacteria isolated from humans (Bifidobacterium infantis, Lactobacillus johnsonii, Lactobacillus rhamnosus, and Lactobacillus acidophilus) animals (Bifidobacterium animalis, 2 strains), and fermented dairy products (Bifidobacterium longum, Lactobacillus kefir, Lactobacillus kefirgranum, and Leuconostoc mesenteroides) with their survival as determined by conventional methods. Five strains were not able to survive (>3 log reduction) 15 min in a medium acidified at pH 2.0 using the conventional testing method, but survival was improved significantly for some strains in the dynamic model. Two strains (Bifidobacterium animalis ATCC 25527 and Lactobacillus johnsonii La-1 NCC 533) showed good survival with both methods. The dynamic model was shown to better represent the events during upper GI tract transit than the conventional methods, by incorporation of a food matrix to buffer the gastric acidity and therefore expose bacteria to pH levels found in vivo before, during, and after a meal.