The objective of the study was to investigate the effects of 2-hydroxy-4-methylthio-butanoic acid (HMTB; Alimet, Novus International, St. Charles, MO) and metabolizable protein, (Low MP or High MP) as observed in vitro and in vivo. Four multiparous, lactating Holstein cows (average DIM 169.5 ± 20.5 d), fitted with a ruminal and duodenal cannula were used in a 4 × 4 Latin square with a 2 × 2 factorial arrangement of treatments. Experimental periods were 28 d in length and cows received a top dress with either 0 or 25 g of HMTBa. On d 23 of each period and 4 h after feeding, 5 kg of whole ruminal contents were collected from all cows and rumen fluid was collected. Inoculum containing rumen fluid from each cow was added to tubes according to treatment and 0.5 g of each dietary treatment placed in a 100 mL fermentation tubes.. After 24 h of fermentation, the fermentation was stop by snap freezing and a microbial pellet was isolated. DNA was isolated from the microbial pellet and was used for microbial community analysis. The Operational Taxonomic Unit(s) (OTU) based analysis of OTU in the Core Measureable Microbiome (CMM), revealed that the relative abundance of Firmicutes decreased (P = 0.01) from the Low MP diet to the High MP diet (41.3 ± 4.9% and 30.2 ± 1.25%, respectively). In contrast, the proportion of Tenericutes, and Spirochaetes increased (P ≤ 0.05) from the Low MP to the High MP diet while an interaction between MP and HMTBa was observed (P ≤ 0.05) in Bacteroidetes, Fibrobacteres, and Verrucomicrobia. The relative abundance of phyla also differed (P ≤ 0.01) by method (in vitro vs in vivo) in Firmicutes, Bacteroidetes, Proteobacteria, TM7, Tenericutes, Spirochaetes, SR1, and Verrucomicrobia. The proportion of Actinobacteria tended to be higher (P = 0.02) in vivo. Comparison of methods revealed differences in the average proportion of most phyla suggesting that in vitro fermentation after 24 h was not completely representative of that which was observed the in vivo.