The ability of Staphylococcus aureus protein A to bind human serum immunoglobulin of different classes and the conditions which allow a selective reduction of IgG in comparison to other immunoglobulin classes were preliminarly studied. Working under standardized conditions, a recovery of 3.5% of the initial value for IgG, of 40-50% for IgM and of 70-75% for IgA was obtained. Rubella HI antibodies, before and after staphylococcus treatment, were titred in 88 sera collected from adult women at different times from hexantema and in 210 sera of 109 subjects exposed to eventual contamination. The results were compared with the ones obtained by means of sucrose density gradient centrifugation followed by titration of rubella HI antibodies in the fractions. The tests proved that HI antibodies titration, before and after serum treatment with staphylococcal protein A, can be used, as presumptive test, for the research of rubella specific IgM antibodies. In fact, all the sera, on which the presence of rubella IgM antibodies was proved by sucrose density gradient centrifugation, were also positive after staphylococcus treatment. In 15% of sera resulting positive for HI rubella antibodies after staphylococcus treatment, usually at low titer, the presence of specific IgM antibodies was not detected by serum fractionation on sucrose density gradient ("false positive" cases). The results of serological diagnosis, carried out by using treatment with staphylococcal protein A and serum fractionation on sucrose density gradient, matched the clinical-epidemiological data. Twenty-seven women, out of 28 examined during the first three months of pregnancy and for whom serological diagnosis of a "recent" rubella virus infection was negative, had normal children at normal time. Therefore only in one case a spontaneous abortion occurred for uncertain causes. Rubella virus was isolated from the embryo of a woman who interrupted pregnancy because asymptomatic rubella virus infection was serologically diagnosed in the second month.