Nearly full-length genomic segments 2 and a partial-length genomic segment 1 of human picobirnavirus were cloned and sequenced. The clones were derived from viruses obtained from human immunodeficiency virus (HIV)-infected patients in Atlanta, Georgia (strains 3-GA-91 and 4-GA-91) and a nonHIV-infected person from China (strain 1-CHN-97). The picobirnavirus genomic segments lacked sequence similarities with other viral sequences in GenBank and EMBL. Comparison of genomic segment 1 from a human and a rabbit picobirnavirus identified a region of 127 nucleotides with 54.7% identity. The genomic segments 2 of the 4-GA-91 and 1-CHN-97 strains had 41.4% nucleic acid identity and 30.0% amino acid similarity and contained amino acid motifs typical of RNA-dependent RNA polymerase genes. Reverse transcription-PCR detection assays were developed with primers targeted to the genomic segments 2 of strains 4-GA-91 or 1-CHN-97. Picobirnaviruses related to the China strain were the predominant viruses detected in stool samples from people in four countries on three continents. Picobirnaviruses were detected in samples from two outbreaks of gastroenteritis in long-term elder care facilities but were not determined to be the primary pathogen. Our findings support the view that picobirnaviruses constitute a distinct family of viruses.