Microbial population in the gastrointestinal tract plays a central role in health and nutrient digestion. The objective of the present study was to investigate the relationships between microbiota and apparent digestibility coefficients with respect to age and diet. Pigs from Large-White, Duroc or Pietrain breeds were raised under the same housing conditions and fed alternately a low-fiber (LF) and a high-fiber diet (HF) during 4 successive 3-week periods. Data collection for digestibility measurements was achieved during the last week of each period. At the end of each period, fecal microbiota was collected for 16S rRNA gene sequencing. The microbiota remained stable across periods whereas digestibility of energy, crude proteins and cell wall components increased. The microbiota was resilient to diet effect and pigs fed the LF diet were discriminated to those fed the HF diet using 31 predicting OTUs with a mean classification error-rate of 3.9%. Clostridiaceae and Turicibacter were negatively correlated whereas Lactobacillus was positively correlated with protein and energy digestibility coefficients in the LF group. In addition, Lachnospiraceae and Prevotella were negatively correlated with cell wall component digestibility. In contrast, no significant correlation was found between microbiota composition and digestibility coefficients when pigs were fed the HF diet. Interestingly, it was also no longer possible to distinguish animals from different breeds once the animals were fed a HF diet, so that the microbiota could only trace the breed origin in the first period and in the LF group. In our experimental conditions, 3 weeks of adaptation to a new diet seems to be sufficient to observe resilience in growing pigs' microbiota. We demonstrated that fecal microbiota can be used to classify pigs according to their dietary treatment. Some bacteria are favorable or unfavorable to digestibility. This suggests that manipulations of bacterial populations can improve digestibility and feed efficiency.