Dietary effects on the intestinal microflora have gained increasing interest because of the evidence that a balanced micro ecology in the gut is important for health and well being. The aim of the present study was to evaluate the effect of different diets on faecal counts of bifidobacteria and Clostridium perfringens in dogs. Two extruded, dry diets, one supplemented with 3% chicory (1.5% inulin), a non-digestible oligosaccharide (NDO) and the other with 3% glucose (GLU) were compared with a protein rich diet (PR+) based on low quality animal derived protein sources (NDO 265, GLU 259, PR+ 726 g crude protein/kg dry matter; greaves meal and bovine lung as protein sources in PR+). Nine adult beagles were subjected to a consecutive cross-over trial. All dogs started with diet PR+, after which groups of four dogs (group A) received GLU and the other five dogs (group B) received NDO. After an intermediate wash-out period with diet PR+ for 3 weeks the A dogs were switched to diet NDO and B dogs to GLU. In the final period all dogs were fed with diet PR+. Faecal samples were collected during each period for dry matter and pH measurements. Faecal bifidobacteria and Cl. perfringens were quantified in fresh samples at the end of each feeding period and additionally on the first days after feed change from the dry diets to diet PR+. Diets NDO and GLU increased faecal dry matter and reduced faecal pH from 6.9 to 7.4 with the high protein diet to 5.9-6.5. The dry diets induced a firmer faecal consistency and a lower faecal pH, with no significant difference between NDO or GLU. Clostridium perfringens was found in all faecal specimens after feeding PR+ with counts of log 8.2-8.8 colony forming units (cfu)/g faeces. Both dry diets reduced the counts of Cl. perfringens significantly (log 3.3-4.0 cfu/g faeces). Switching from the dry diets to the high protein diet induced an increase of Cl. perfringens within 1 day, independent of the previous diet. In dogs fed PR+, bifidobacteria were detected in only four faecal samples and exclusively in the initial feeding period. During the remainder of the experiment the counts fell below the detection limit (log 6 cfu/g faeces). The faecal concentrations of bifidobacteria increased with both dry diets. Slightly higher concentrations (log 9.6-9.7 cfu/g faeces) were obtained from dogs fed the dry diet containing NDO compared with the diet containing glucose (log 9.3-9.4 cfu/g faeces). The increase was small which may be related to the level of total fermentable carbohydrates in both diets which alone increase remarkably the total counts of bifidobacteria. In conclusion, distinct dietary effects on the faecal counts of Cl. perfringens and bifidobacteria with a clear antagonistic pattern were observed. The main factor was the protein source and level in the diet. In this case, NDO favoured the concentrations of bifidobacteria to a limited degree. Further studies are needed to evaluate time effects, metabolic consequences and the potential implication for health promotion in pets.