Plastid genomes of flowering plants are largely identical in gene order and content, but a few lineages have been identified with many gene and intron losses, genomic rearrangements, and accelerated rates of nucleotide substitutions. These aberrant lineages present an opportunity to understand the modes of selection acting on these genomes as well as their long-term stability. My research has focused on two areas within plastid genome evolution in Geraniaceae: first, an investigation of the diversity of unusual plastid genomes in a single genus, Erodium (Geraniaceae) for chapters one and three. Chapter two focuses on the evolution of subunits of the plastid-encoded RNA polymerase (PEP). The first chapter described the loss of plastid-encoded NADPH dehydrogenase (ndh) genes from a clade of 13 Erodium species. Divergence time estimates indicate this clade is less than 5 million years old. This recent loss of ndh genes in Erodium presents an opportunity to investigate changes in photosynthetic function through comparative biochemistry between Erodium species with and without plastid-encoded ndh genes. Second, I examined the evolution of the gene encoding the alpha subunit (rpoA) of PEP in three disparate angiosperm lineages—Pelargonium (Geraniaceae), Passiflora (Passifloraceae), and Annonaceae—in which this gene has diverged so greatly that it is barely recognizable. PEP is conserved in the plastid genomes of all photosynthetic angiosperms. I found multiple lines of evidence indicating that the genes remain functional despite retaining only ~30% sequence identity with rpoA genes from outgroups. The genomes containing these divergent rpoA genes have undergone significant rearrangement due to illegitimate recombination and gene conversion, and I hypothesized that these phenomena have also driven the divergence of rpoA. Third, I conducted a survey of plastid genome evolution in Erodium with the completion of 15 additional whole genomes. Except for Erodium and some legumes, all angiosperm plastid genomes share a quadripartite structure with large and small single copy regions (LSC, SSC) and two inverted repeats (IR). I discovered a species of Erodium that has re-formed a large inverted repeat. Demonstrating a precedent for loss and regain of the IR also impacts models of evolution for other highly rearranged plastid genomes. / text