The human genome contains a large number of sequences that belong to the HERV-K family of human endogenous retroviruses. Most of these elements are likely remnants of ancient infections by ancestral exogenous retroviruses. To obtain further insight into the evolutionary history and molecular mechanisms responsible for the diversity of the human HERV-K elements, we analyzed several aspects of their genome structure. The nucleotide composition of the HERV-K genome was found to be highly biased and asymmetric, with an abundance of the A nucleotide in the viral (+) strand. A similar trend has been reported for the genomes of several exogenous retroviruses, with different nucleotides as the preferred building block. Other genome characteristics that were reported previously for actively replicating retroviruses are also apparent for the endogenous HERV-K virus. In particular, we observed suppression of the dinucleotide CpG, which represents potential methylation sites, and a strong preference for synonymous substitutions within the open reading frame of the reverse transcriptase (RT) enzyme. Furthermore, the mutational spectrum of the HERV-K RT enzyme was evaluated by nucleotide sequence comparison of 34 available elements. Interestingly, this analysis revealed a striking similarity with the mutational pattern of the HIV-1 RT enzyme, with a preference for G-to-A and C-to-T transitions. It is proposed that the mutational bias of the HERV-K RT enzyme played a role in the shaping of this retroviral genome, which was actively replicating more than 30 million years ago. This effect can still be observed in the contemporary endogenous HERV-K elements.