OBJECTIVE Ranges of variation and conservation in sequence need to be defined for detecting and genotyping hepatitis E virus (HEV). METHODS Six HEV isolates from Japanese patients were sequenced over the entire genome and compared phylogenetically along with 16 reported HEV isolates, including two from pigs. RESULTS Three of the six HEV isolates were of genotype III, and the remaining three were of genotype IV. Local clusterings of Japanese HEV isolates were observed in the phylogenetic analyses, including a swine HEV isolate reported previously (swJ570). All six HEV isolates possessed three open reading frames (ORFs). The ORF3 in the three isolates of genotype III were in a different reading frame, while that in the three isolates of genotype IV were in the same reading frame as ORF1. A stretch of 46-96 nucleotides was identified, point mutations and deletions in which were specific for the four genotypes (I-IV). A polymerase chain reaction method was developed with 9 nested universal primers, deduced from conserved regions in the 5'-terminal sequences of the 22 HEV genomes. CONCLUSIONS Conserved and genotype-specific variation in HEV sequences, identified in the comparison of 22 full-length genomes, would be useful in designing primers for sensitive detection and specific genotyping of HEV RNA.