Red algal plastid genomes are often considered ancestral and evolutionarily stable, and thus more closely resembling the last common ancestral plastid genome of all photosynthetic eukaryotes [ 1 , 2 ]. However, sampling of red algal diversity is still quite limited (e.g., [ 2 , 3 , 4 , 5 ]). We aimed to remedy this problem. To this end, we sequenced six new plastid genomes from four undersampled and phylogenetically disparate red algal classes (Porphyridiophyceae, Stylonematophyceae, Compsopogonophyceae, and Rhodellophyceae) and discovered an unprecedented degree of genomic diversity among them. These genomes are rich in introns, enlarged intergenic regions, and transposable elements (in the rhodellophycean Bulboplastis apyrenoidosa ), and include the largest and most intron-rich plastid genomes ever sequenced (that of the rhodellophycean Corynoplastis japonica ; 1.13 Mbp). Sophisticated phylogenetic analyses accounting for compositional heterogeneity show that these four “basal” red algal classes form a larger monophyletic group, Proteorhodophytina subphylum nov., and confidently resolve the large-scale relationships in the Rhodophyta. Our analyses also suggest that secondary red plastids originated before the diversification of all mesophilic red algae. Our genomic survey has challenged the current paradigmatic view of red algal plastid genomes as “living fossils” [ 1 , 2 , 6 ] by revealing an astonishing degree of divergence in size, organization, and non-coding DNA content. A closer look at red algae shows that they comprise the most ancestral (e.g., [ 2 , 7 , 8 ]) as well as some of the most divergent plastid genomes known.