We propose to classify "genome types" of human adenoviruses in the same way as the "genomic clusters" defined by Li and Wadell, i.e. groups of closely related DNA variants of a given serotype named by the use of indices, P (standing for prototype) a, b etc. Variants within the genome type are named by letter-number indices (p1, p2; a1, a2 etc.). The percentage of comigrating restriction fragments and common restriction sites is significantly higher in variants than between genome types. The new naming system is applicable to adenoviruses from 5 subgenera, but requires additional work for subgenus C strains. Renaming is required mainly for adenovirus 7.