In the present study serum and mucosal antibody responses to the M protein of Streptococcus equi subsp. equi were studied in a group of young horses following intranasal application of an avirulent strain of subsp. equi and during a subsequent experimentally induced strangles epizootic in an attempt to identify humoral immune responses that could be correlated with protection. Convalescent protective immunity was verified by intranasal inoculation of virulent subsp. equi 5 weeks after recovery. Additionally, epizootiological and clinical data were collected during the strangles epizootic to gain further insight into the epizootiology and dynamics of subsp. equi infection within the group. Serum IgG and mucosal IgA and IgG levels were measured by ELISA with electrophoretically purified acid extracted fragments of the M protein of subsp. equi. ELISA values were set in relation to total IgA and IgG antibody to determine M protein specific local IgA and IgG activities in nasal wash samples. Serum bactericidal responses were measured by counts of viable organisms after opsonization and incubation with horse neutrophils for 3 hours. Intranasal inoculation of 12 horses with an avirulent strain of subsp. equi did not elicit protective immune reponses. Following commingling exposure to 2 experimentally infected horses all 20 horses (12 inoculated and 8 controls) developed strangles of varying severity with a mean incubation period of 12 days. The horses were not medicated during the epizootic. Thirteen horses challenged by intranasal inoculation 5 weeks after recovery were solidly immune. At this time these horses exhibited strong serum bactericidal activity and high levels of M protein specific serum IgG. Additionally, nine horses showed high levels of M protein specific mucosal IgA at the time of challenge, whereas local IgG had already returned to preinfection values in most horses. Intranasal challenge did not produce a local or systemic booster response, indicating that protective antibodies may have prevented immune stimulation by the antigens presented at this time. A striking feature of the study was the great qualitative and quantitative heterogeneity of serum and mucosal antibody response among individual horses. Systemic and local antibody responses were not synchronized, suggesting independence of these responses. A substantial portion of M protein reactive serum IgG showed crossreactivity with S. equi subsp. zooepidemicus, suggesting extensive sharing of epitopes by the two subspecies. Local IgA concentrations were about 10 times higher than local IgG concentrations. IgG in nasal wash samples exhibited high M protein specific activity, suggesting local synthesis of this antibody. Nasal shedding of subsp. equi as studied intensively in 9 horses started after a latent period of 2-9 days and ceased after a mean of 14 days. Shedding by 2 horses continued for 61 and 106 days after clinical signs were first noticed. Both horses showed intermittent purulent nasal discharge until about 3 weeks before the last isolation of subsp. equi. Cessation of shedding was associated in most horses with appearance of high levels of M protein specific local IgA. Barn repopulation with a highly susceptible group of yearlings 8 months after the strangles epizootic was uneventful, despite the lack of preceding decontamination measures such as barn disinfection.