Ruminal acidosis is a recurrent problem in ruminant production systems. The prevailing fermentation acid is difficult to predict in practice and it is thought to be the reason for the variable effectiveness of some rumen modifiers. We developed a sheep model of experimentally - induced lactic, butyric or propionic acidosis to describe the associated changes in the rumen microbiota. Twelve rumen - cannulated sheep separated into 3 gr oups of 4 animals were fed a hay:concentrate diet (80:20) during 3 weeks followed by three consecutive days of acidosis challenge by direct intraruminal supply (1.2% BW) of ground wheat, corn or beet pulp. Rumen contents were analysed for fermentation para meters (rumen pH, volatile fatty acids, lactate) and polysaccharidase activities (amylase, xylanase, carboxymethyl - cellulase). Populations of Fibrobacter succinogenes,Ruminococcus albus, ruminococcus flavefaciens, Streptococcus bovis, Lactobacilli, Prevote lla and total bacteria were quantified by qPCR and protozoa were counted under photonic microscope. At the end of the challenge, rumen pH was lower for wheat (4.87) than for corn (5.17) and beet pulp (5.55). Lactate was predominant for wheat (33.8 mM vs. ~3 mM for other challenges), whereas beet pulp and corn favoured a high molar percentage of propionate (22.5%) and butyrate (16.9%), respectively. R. albus and F. succinogenes proportions were highest for wheat than corn and beet pulp while R. flavefaciens proportions and fibrolytic activities, total bacteria and protozoa were similar among treatments. Amylase activity was highest for wheat, intermediate for corn and lowest for beet pulp. Lactobacilli and S. bovis proportions tended to be highest for wheat (1.74 vs. ~0.12%; P = 0.15) and corn (0.49 vs. ~0.05%; P = 0.17) respectively, whereas Prevotella was highest for beet pulp (49 vs. ~29%). This study demonstrated that, feed - induced lactic, butyric or propionic acidosis were associated with specific change s in the bacterial population.