The immunobiology underlying the slow acquisition of immunity to group A streptococci (GAS), is not understood, but attributed to specific virulence factors impeding innate immunity and significant antigenic diversity of the type-specific M-protein, hindering acquired immunity. This results in extreme endemicity and very high rates of streptococcal-associated serious pathology, including rheumatic heart disease, post streptococcal glomerulonephritis and invasive GAS disease. We used a number of epidemiologically distinct GAS strains to model the development of acquired immunity to pyoderma. We show that infection leads to antibody responses to the serotype-specific determinants on the M protein and profound protective immunity; however, memory B cells do not develop and immunity is rapidly lost. Two sequential infections with the same strain within a short time frame were required to induce enduring strain-specific immunity. Sequential infections with different strains resulted in partial immunity only to the last strain to which they had been exposed and not to any previous strains. Mice exposed to multiple strains, either sequentially or simultaneously, did not develop antibodies to a conserved M protein vaccine peptide, J8, demonstrating that this epitope is cryptic to the immune system. However, in contrast to the lack of strain-specific immunity that follows infection, immunity following vaccination with J8 protects against multiple strains delivered sequentially or as a co-infection. Moreover, vaccine-induced immunity could be boosted by sequential heterologous infections. This highlights a major difference in how naïve and memory B cells respond to cryptic epitopes which could have important implications for vaccine research.