Polymerase chain reaction (PCR)-based denaturation gradient gel electrophoresis (DGGE) is currently being used for characterizing the composition of the gut microbiota (GM) of mice in order to better control the study variation arising from the GM. At present, faeces are commonly sampled from live animals, while caecum is most commonly sampled from terminated animals. However, there is no knowledge whether the composition at the one site is representative for the other. In this study C57BL/6 mice were observed from the age of four weeks until the age of 10 weeks. Faeces were sampled weekly. Caecum was sampled surgically under anaesthesia and with subsequent ampicillin treatment at the age of six weeks and again after euthanasia at the age of 10 weeks. Faecal and caecal microbiota profiles were determined using DGGE and subjected to subsequent cluster analysis. The mice subjected to surgical caecal sampling clustered separately for two weeks after termination of antibiotics after which they again clustered with the non-surgically sampled mice. Faecal and caecal profiles clustered separately at the age of six weeks, but not at the age of 10 weeks. There were no correlations between faecal or caecal profiles at six or 10 weeks of age, respectively. It is concluded that faecal and caecal microbiota profiles are not representative of each other in mice. Therefore, it is recommendable in studies to sample from several sites specifically decided in relation to the specific model of a study.