It has been 22 years since the first human cytomegalovirus (HCMV) full genome sequence was published. Together with the first sequence also came the first prediction of the coding capacity of this intriguing herpesvirus, harbouring the largest genome of all human-infecting viruses. Through the years, the HCMV genome map was refined, but it is still far from complete. Comparisons of the genomes of highly passaged and attenuated laboratory strains with those of recent clinical isolates have shown the former underwent substantial rearrangements with gene deletions and duplications. It became clear that different clinical isolates display remarkable genetic variability in large portions of their genomes. This led to a quest for clinical correlates of different genetic variants. Until now, this approach did not give satisfactory results and it is still not clear if certain genetic variants are associated with distinct disease outcome. More recently, extra levels of complexity have been added to the existing picture of HCMV genetic variability. Infections with multiple HCMV strains seem to be very common and the dynamics of different strains show unpredictable and stochastic behaviour. Deep sequencing efforts have unveiled an extensive intrapatient genetic variability, which was unexpected for a dsDNA virus. Recent efforts to elucidate the HCMV transcriptome have painted a sophisticated picture with lots of splicing and alternative transcripts. It is clear that we are only unveiling the tip of the iceberg regarding HCMV genomics and transcriptomics and their role in the myriad viral functions. Recent advances in technology and bioinformatics provide us with the tools to begin to tackle these interesting questions.