The more we learn about our immune system, the more amazed we are by its complexity. Just when immunologists thought the basic principles for the somatic diversification of antigen receptors for B and T cells had been solved, we began to appreciate the complexity of the innate immune response, its importance as a first level of protective immunity and its role in triggering antigenspecific responses by the adaptive immune system. Perhaps if we could work out the evolutionary history of innate and adaptive immunity, this could help us to understand the intricacies of how our immune system functions and sometimes dysfunctions. Innate immune mechanisms can be found in representative species at almost every level of the evolutionary tree of life. This fact alone is evidence of the importance of innate immunity in the competitive ‘struggle for existence’1 that began with the appearance of single cell microorganisms on Earth more than 3.5 billion years ago. The ensuing evolution of diverse bacteria, archaea and eukaryotes was so successful that these microorganisms caused environmental changes, including an increase in the concentration of atmospheric oxygen, that fostered the development of multicellular organisms (metazoans) around 600 million years ago. The evolutionary burst of diversity in metazoan species over the next 50 million years presented new host opportunities for microbial pathogens. In turn, the need for new mechanisms of host defence might explain the remarkable diversity of innate defence mechanisms in plants and animals. Although many different innate immune mechanisms are deployed for host defence, a unifying theme of innate immunity is the use of germlineencoded pattern recognition receptors for pathogens or damaged self components, such as the Toll-like receptors, nucleotide-binding domain leucine-rich repeat (LRR)containing receptors, retinoic acid-inducible gene I-like RNA helicases and C-type lectin receptors2,3. Another layer of complexity in immune defences emerged during chordate evolution with the appearance of adaptive immunity in vertebrates around 500 million years ago. The unique feature of an adaptive immune system is the somatic development of clonally diverse lymphocytes, each of which has a unique antigen recognition receptor that can be used to trigger its activation. The combinatorial generation of a highly diverse lymphocyte receptor repertoire allows vertebrates to recognize almost any potential pathogen or toxin and to mount antigen-specific responses to it. Antigen-activated lymphocytes undergo clonal expansion and differentiation into mature effector lymphocytes with cytotoxic and pro-inflammatory functions or into plasma cells that secrete antibodies (the soluble forms of the antigen-specific receptors). Moreover, the clonal expansion and long life of some antigen-primed cytotoxic lymphocytes and plasma cells provide protective memory to prevent reinvasion. Effective cellular and humoral immune responses by T and B cells, respectively, require the participation of various phagocytic cells, dendritic cells (DCs), natural killer (NK) cells and other types of innate immune cell and humoral components, which have important inductive or effector roles to provide protective immunity4. Tracing the evolution of genes has become easier with the increasing availability of genomic sequences of representative species, but it is more difficult to discern the evolutionary history of the extensive network of individual cell types that must work together for effective immunity. For example, we do not currently know when some of the key immune players entered the evolutionary scene, such as DCs and NK cells. Moreover, continual evolutionary changes add to the confusion. A relevant example is the fairly recent evolution of entirely different types of NK cell receptors in mice and humans, who last shared a common ancestor around 65 million years ago; mouse NK cells use lectin receptors, whereas primate NK cells use immunoglobulin-based killer cell immunoglobulinlike receptors (KIRs) to recognize MHC class I-associated ligands, which control their activation5. As if understanding this complex evolutionary puzzle were not already sufficiently challenging, we have learned recently that two types of adaptive immune system have evolved in vertebrates: a recently recognized system in jawless vertebrates (hagfish and lamprey) and the more familiar adaptive immune system of jawed vertebrates6. These two systems use entirely different types of antigen recognition receptor, but they use similar lymphocyte Max D. Cooper and Brantley R. Herrin are at the Emory Vaccine Center and Department of Pathology and Laboratory Medicine, Emory University, 1462 Clifton Road, Atlanta, Georgia 30322, USA. Correspondence to M.D.C. e‐mail: max.cooper@emory.edu