Diversity in the surface antigens of malaria parasites is generally assumed to be a mechanism for immune evasion, but there is little direct evidence that this leads to evasion of protective immunity. Here we show that alleles of the highly polymorphic merozoite surface protein 2 (MSP-2) can be grouped (within the known dimorphic families) into distinct serogroups; variants within a serogroup show extensive serological crossreactivity. Cross-reactive epitopes are immunodominant, and responses to them may be boosted at the expense of responses to novel epitopes (original antigenic sin). The data imply that immune selection explains only some of the diversity in the msp-2 gene and that MSP-2 vaccines may need to include only a subset of the known variants in order to induce pan-reactive antibodies.Allelic diversity in malaria parasites is widely assumed to be a mechanism for immune evasion (7,24). Among surface proteins, diversity arises from multiple point mutations or variations in the numbers, lengths, and sequences of amino acid repeats. Repeat diversity may arise from intragenic recombination, misalignment of repeated DNA sequences (6), or complementary-strand slippage (12, 13) during DNA replication. The AT-rich nature of the Plasmodium falciparum genome may facilitate the generation of new variants (28). The maintenance of variants in the population implies that they have a selective advantage (e.g., evasion of preexisting immune responses) or that the mutations are selectively neutral and that population diversification occurs randomly.In some genes (e.g., the circumsporozoite surface protein [csp] gene), point mutations cluster in areas immunodominant for T cells (16) and can lead to loss of epitope recognition (38) or antagonism of T-cell responses (26), although it is not clear that either of these leads to the avoidance of protective immunity. Similarly, monospecific antibodies differentiate between allelic variants of merozoite surface proteins (13, 23), but the human antibody response is commonly directed at conserved or semiconserved epitopes (8, 34) and direct evidence for the evasion of antibody-dependent immunity is lacking.Comparison of the frequencies of nonsynonymous-to-synonymous mutations (20) for P. falciparum genes provides evidence of diversifying selection in msp-1 (11), which encodes merozoite surface protein-1, but the data are less clear for msp-2, which encodes merozoite surface protein-2 (9, 11, 21), and the significance of the findings is uncertain, given the limitations of the method (37). Comparison of intra-and interpopulation variances in allele frequencies provides evidence for selection on a polymorphic region of msp-1 (5) and dimorphic regions of msp-2 (4), but there are no data for polymorphic regions of msp-2. Comparison of inter-and intraspecific levels of variation suggests balancing selection on the erythrocyte-binding antigen EBA-175 (25) and the apical membrane antigen AMA-1 (22). However, except for msp-1, for which selection may be mediated by strain-specific antibo...