Rotavirus A (RVA) causes acute gastroenteritis in humans and animals globally, killing 450,000 children per year, mainly in developing countries (1).The RVA genome includes 11 double-stranded RNA segments encoding six viral proteins (VP1 to VP4, VP6, VP7) and six nonstructural proteins (NSP1 to NSP6) (2). Rotavirus outer layer protein VP7 and spike protein VP4 are neutralization antigens, determining virus genotypes G (glycoprotein; VP7) and P (protease sensitive; VP4). At least 27 G and 37 P genotypes have been identified among human, animal, and avian RVAs (3-5).Five genotypes ( Serotypes other than G1 to G4 and strains evolving by mutation or genomic reassortment between human and/or animal RVAs represent an emerging threat to children (8-10). Millions of people are exposed to animal rotaviruses worldwide, and several uncommon human genotypes (G6, G8, G12, P3[9], P5A[3]) closely resemble animal RVA strains (11).Molecular epidemiology and viral typing can help monitor the emergence of novel RVAs, confirming the efficacy of current vaccines against unconventional genotypes.Reassortment events can involve any double-stranded RNA segments, and G and P typing may be insufficient to investigate RVA origin and interspecies transmission. Genotyping based on sequencing of all 11 genome segments (3, 4) is very helpful, distinguishing 9 VP1 (R), 9 VP2 (C), 8 VP3 (M), 35 VP4 (P), 16 VP6 (I), 27 VP7 (G), 16 NSP1 (A), 9 NSP2 (N), 12 NSP3 (T), 14 NSP4 (E), and 11 NSP5 (H) genotypes.Bovine-like G8 rotavirus was first reported in an Indonesian child (12), and further G8 RVA human cases occurred worldwide, including in industrialized countries (13-21). Using full-genome sequencing, G8P[8] African RVAs were shown to involve reassortment between at least four human, swine, and bovine strains (22). This paper reports the whole-genome characterization of a G8P [8] human rotavirus strain that emerged in Croatia in 2006 (23).Cases showed severe gastroenteritis and included children Ͻ5 23). Agarose gel-purified amplicons were sequenced using 18,24,25), assembling consensus sequences for all gene segments (18,22,26). Twelve data sets were built and included all NCBI sequences (http://www.ncbi.nlm.nih.gov/pubmed) showing Ͼ89% similarity with query sequences (between 116 for VP2 and 127 for NSP4).Sequences were aligned using ClustalX software (27) and edited by Bioedit software. The ModelTest v3.0 (28) with the hierarchical likelihood ratio test was used to select the best-fit models for sequence data analysis.Bayesian phylogenetic trees were constructed for the VP7, VP4, and NSP4 data sets (29), using the GTRϩIϩG nucleotide substitution model for VP7 and the HKYϩG model for VP4/NSP4.Markov chain Monte Carlo searches were made (50 ϫ 10 6 generations, tree sampling every 5,000 generation, and 10% burn-in fraction; clade statistical support followed Ͼ0.90 posterior probability). Maximum-likelihood (ML) phylogenetic trees of NSP1 to NSP3, NSP5 and NSP6, VP1 to VP3, and VP6 were generated with the PAUP* v4.0 package (30), using the GTRϩIϩ...