The objective of this study was to compare the in vivo effect of lactulose on faecal parameters with the effect in a dynamic, computer-controlled in vitro model of the proximal large intestine (TIM-2). Faecal samples from 10 human volunteers collected before (non-adapted) and after 1 week of treatment (10 g/day) with lactulose (lactulose-adapted) were investigated. Parameters were compared immediately in the faecal samples, and after incubation in the in vitro model of the large intestine. After an adaptation period of the faecal microbiota in the in vitro model of the proximal colon, lactulose (10 g/day) was fed to the microbiota over a 48-h period. Samples taken from the model were investigated for microbiota composition and metabolite production (short-chain fatty acids (SCFAs) and lactate). No changes in the faecal parameters pH, dry weight or SCFA ratio were observed in the in vivo samples. However, the results show a major change in the ratio of SCFAs produced in the in vitro model, with a drastic reduction of butyrate production on lactulose. This was clear in the non-adapted microbiota by the observed arrest in butyrate production 24 h after the start of lactulose feeding. However, in the adapted microbiota butyrate production was already low from the start of the experiment. In fact, only the microbiota of one of the 10 individuals still produced significant amounts of butyrate after lactulose adaptation, the concentration in the other samples was extremely low. Similarly, in the in vitro model lactate production of the non-adapted microbiota started after approximately 24 h, whereas the adapted microbiota produced lactate from the start. In faecal (in vivo) samples no changes in microbiota composition were obvious, except for a significant increase in Bifidobacterium counts after lactulose feeding. With classic plating techniques, the in vitro samples showed an increase in Lactobacillus and Enterococcus species. With denaturing gradient gel electrophoresis, a clear change in banding pattern was observed, indicating a shift in microbiota composition. When the major bands that appeared after lactulose feeding in the in vitro model were excised and sequenced, the sequences showed homology to Lactobacillus and Enterococcus species. This is in agreement with the classic plating technique as well as with the observed increase in lactate production. Sampling in vivo at 'the site where it all happens' (the proximal colon) is difficult and inconvenient. We conclude that the in vitro model for the proximal colon reflects much better the fermentation of lactulose, in both metabolite production and changes in microbiota composition, than do faecal samples from an in vivo experiment. Therefore, the in vitro model is an excellent tool with which to study bioconversion of functional food components and/or drugs.