The human adenoviruses are classified according to their nucleotide sequence homology and their oncogenic potential in rodents. The left-hand end of the genome of the adenovirus types 5 (ad5), 7 (ad7) and 12 (ad12), which are respectively, non-oncogenic (subgroup C), weakly oncogenic (subgroup B) and highly oncogenic (subgroup A), contains all the genetic information needed to induce and maintain the transformed phenotype. This part of the genome contains the early transcription unit designated E1 which is subdivided in two transcription units E1A and E1B. Two spliced mRNAs are transcribed from the E1A region which codes for several phosphorylated polypeptides. These polypeptides play a key role by controlling the expression of the other early transcription units. The major role of region E1A in adenovirus cell transformation might not be activate the true transforming genes of the region E1B. An additional role probably consists of the activation of some cellular genes as a restriction fragment containing this region can immortalize rodent cells in vitro. An important question is why some adenoviruses are oncogenic and others are not. We report here differences in the structures of the E1A polypeptides from ad7 and ad12, compared to ad5, which may partially account for their differing oncogenicity.