IgA levels in saliva differed 15-fold between different individuals. The correlation between IgA in repeated stimulated saliva samples was good in both a short and a 2-yr period. The correlation between resting and stimulated samples was low, in particular when output of IgA over time was considered. Physical exercise affected salivary protein concentrations and also percentage of IgA to protein. The elevated IgA levels in saliva at pregnancy were not accompanied by a similar change of IgA activity against specific bacteria.