Studies using fistulated ruminants have yielded tools to determine the feed efficiency of diets, but their ethical acceptability is put into question. Our objective was to determine whether oral-stomach sampling (OSS) is an acceptable alternative to sampling through a ruminal cannula in characterizing the variability of rumen fluid composition induced by an acidogenic dietary challenge in dairy cows. During three four-week periods (P1, P2, P3), six rumen-fistulated cows were fed a standard diet based on supplemented corn silage (periods 1 and 3) or a diet enriched with starch (period 2). Rumen fluid was collected through cannula at three locations (reticulum, ventral sac or a mix of both) and by OSS, once per week at 0830 h (before morning feeding), and once every third week of each period at 1330 h. Ruminal pH was higher when obtained by OSS rather than cannulation, whatever the sampling location (+0.44 points at 0830 h and +0.56 points at 1330 h, P < 0.001), whereas ruminal concentrations of VFAs were lower and molar proportions of VFAs were either unaffected by the sampling method/ location (butyrate, iso-VFAs, P > 0.10) or affected but with a limited magnitude of variation (P < 0.05). Our experimental design did not allow the detection of a difference in ruminal VFA concentration between acidogenic and control conditions before the morning meal with OSS whereas it was detected with cannula sampling. Our results illustrates that the procedure and recommendations for the use OSS in experimental design still need to be refined.