Recent twin studies of clinical malaria and immune responses to malaria antigens have underscored the importance of both major histocompatability complex (MHC) and non-MHC genes in determining variable susceptibility and immune responsiveness. By using a combination of whole genome genetic linkage studies of families and candidate genes analysis, non-MHC genes are being mapped and identified. Human leucocyte antigen (HLA) genotype was found to affect susceptibility to severe malaria in a large study of West African children. T lymphocytes that may mediate such resistance have been identified and their target antigens and epitopes characterized. Some of these epitopes show substantial polymorphism, which appears to result from immune selection pressure. Natural variant epitopes have been found to escape T-cell recognition in cytolytic and other T-cell assays. More recently a novel immune escape mechanism has been described in viral infections, altered peptide ligand antagonism, whereby variants of a T-cell epitope can downregulate or ablate a T cell response to the index peptide. The likely implications of such immune escape mechanisms for the population structure of malaria parasites, for HLA associations with malaria infection and disease, and for the design of new malaria vaccines, are discussed. The evolutionary consequences of such molecular interactions can be assessed by using mathematical models that capture the dynamic of variable host and parasite molecules. Combined genetic, immunological and mathematical analysis of host and parasite variants in natural populations can identify some mechanisms driving host-parasite coevolution.