A model for the molecular evolution of an imune system is presented. It suggests how a system of cell surface-fixed antigen receptors, called FR, to be thought of as "primitive" but still functional antibodies, could evolve into the "modern" labile T cell and antibody system, the receptors for which may be called CER (clonal expansion receptors). Perhaps the most significant insight to be gained from the theory concerns the conclusion that the immune system of an embryo may be primed by the interaction between two types of cells with complementary surface specificities. In the first type of cell the interacting molecule is an FR molecule coded by a gene of the Major Histocompatibility Complex (MHC) which has undergone a somatic mutation; in the second type it is a CER molecule also coded by a gene which has undergone a recent somatic mutation. It is believed that this insight eliminates some of the problems experienced in understanding generation of diversity, and renders unnecessary some of the more complex hypotheses about immune networks. The mechanism of action of immune response genes is easily explained by the theory.