Exercise has a significant effect on different physiological systems in the body of human and animals. Only limited numbers of published studies in laboratory animals or humans have shown the effect of exercise on the gut microbiota, and no studies have shown this effect in horses. In this study, 8 horses (4 mares, 4 geldings) were exercise trained for 12 weeks, and 4 additional mares were used as a parallel seasonal control. To identify bacterial community changes over time for both groups, rectal faecal samples were collected, DNA was extracted, and the 16S rRNA gene (V3-V4) was sequenced using the Illumina Miseq platform. One-way ANOVA, Shannon diversity index, and Principal Coordinate Analysis (PCoA) were used to identify differences between and among samples. The exercise training group showed significant changes in the levels of Bacteroidetes, Proteobacteria, and Spirochaetes phyla (P<0.05), while there were no changes in the gut microbiota of the seasonal control group through the three months of the study (P>0.05). Moreover, with training two genera significantly changed in their relative abundance over time, namely Clostridium and Dysgonomonas (P<0.05). Dysgonomonas spp. was significantly changed in abundance during the exercise training period (P<0.05). Also Treponema spp. showed significant changes during the exercise training period (P<0.05). Shannon diversity index was decreased (P<0.05) in the exercise group at the beginning of the study, but then returned to pre-training levels. PCoA showed significant separation between time points of the exercise training group as far as the levels of genera and species (P<0.05) represented. Our results show that exercise training influences the gut microbiota, especially at the beginning of training.