Mutations in the human leucine-rich repeat kinase 2 (LRRK2) gene are associated with both familial and sporadic Parkinson disease (PD). LRRK2 belongs to a gene family known as Roco. Roco genes encode for large proteins with several protein domains. Particularly, all Roco proteins have a characteristic GTPase domain, named Roc, plus a domain of unknown function called COR. In addition, LRRK2 and several other Roco proteins also contain a protein kinase domain. In this study, I use a combination of phylogenetic and structural analyses of the COR, Roc, and kinase domains present in Roco proteins to describe the origin and evolutionary history of LRRK2. Phylogenetic analyses using these domains demonstrate that LRRK2 emerged from a duplication that occurred after the protostome-deuterostome split. The duplication was followed by the acquisition by LRRK2 proteins of a specific type of N-terminal repeat, described here for the first time. This repeat is absent in the proteins encoded by the paralogs of LRRK2, called LRRK1 or in protostome LRRK proteins. These results suggest that Drosophila or Caenorhabditis LRRK genes may not be good models to understand human LRRK2 function. Genes in the slime mold Dictyostelium discoideum with structures very similar to those found in animal LRRK genes, including the protein kinase domain, have been described. However, phylogenetic analyses suggest that this structural similarity is due to independent acquisitions of distantly related protein kinase domains. Finally, I confirm in an extensive sequence analysis that the Roc GTPase domain is related but still substantially different from small GTPases, such as Rab, Ras, or Rho. Modeling based on known kinase structures suggests that mutations in LRRK2 that cause familiar PD may alter the local 3-dimensional folding of the LRRK2 protein without affecting its overall structure.