In the 2010s, several unusual rotavirus strains emerged, causing epidemics worldwide. This study reports a comprehensive molecular epidemiological study of rotaviruses in Japan based on full-genome analysis. From 2014 to 2019, a total of 489 rotavirus-positive stool specimens were identified, and the associated viral genomes were analyzed by next-generation sequencing. The genotype constellations of those strains were classified into nine patterns (G1P[8] (Wa), G1P[8]-E2, G1P[8] (DS-1), G2P[4] (DS-1), G3P[8] (Wa), G3P[8] (DS-1), G8P[8] (DS-1), G9P[8] (Wa), and G9P[8]-E2). The major prevalent genotype differed by year, comprising G8P[8] (DS-1) (37% of that year’s isolates) in 2014, G1P[8] (DS-1) (65%) in 2015, G9P[8] (Wa) (72%) in 2016, G3P[8] (DS-1) (66%) in 2017, G1P[8]-E2 (53%) in 2018, and G9P[8] (Wa) (26%) in 2019. The G1P[8]-E2 strains (G1-P[8]-I1-R1-C1-M1-A1-N1-T1-E2-H1) isolated from a total of 42 specimens in discontinuous years (2015 and 2018), which were the newly-emerged NSP4 mono-reassortant strains. Based on the results of the Bayesian evolutionary analyses, G1P[8]-E2 and G9P[8]-E2 were hypothesized to have been generated from distinct independent inter-genogroup reassortment events. The G1 strains detected in this study were classified into multiple clusters, depending on the year of detection. A comparison of the predicted amino acid sequences of the VP7 epitopes revealed that the G1 strains detected in different years encoded VP7 epitopes harboring distinct mutations. These mutations may be responsible for immune escape and annual changes in the prevalent strains.