A solid-phase antigen enzyme-linked immunosorbent assay (ELISA) was developed for the detection of rubella-specific IgG subclasses. For rubella-specific IgG1 and IgG3 sera were quantitated in arbitrary units (au) by comparison with standard curves. A concentration of 3 au was taken as that indicating positivity for specific IgG1 and specific IgG3. No sera reactive for specific IgG2 and IgG4 have been found, and thus the assay reagents were controlled by testing dilutions of a standard calibrant serum containing known concentrations of the specific IgG subclasses. Of 105 unselected sera negative for rubella antibody by radial haemolysis (RH), two gave concentrations of specific IgG1 greater than 3 au and both were positive by rubella latex agglutination (LA). The sensitivity of the assay for specific IgG1 was confirmed by examining 25 selected sera negative by RH but reactive by LA. Twenty-one gave concentrations greater than 3 au. None of these 130 was positive for specific IgG3. All 63 sera containing greater than 15 international units rubella antibody by RH from cases of rubella in the remote past contained specific IgG1 and eight contained specific IgG3. In 79 cases of primary rubella, specific IgG1 developed in all cases by day 8. Specific IgG3 became detectable in all cases except one by day 16. Serum taken on day 21 from one case was negative for specific IgG3 but the absence of later sera precluded further investigation. One case had become negative for specific IgG3 by day 56. Sera from 24 cases of rubella reinfection were examined and all contained specific IgG1. In three cases of symptomatic reinfection, specific IgG3 was detectable in two but not in the remaining case. In 2 of the 21 cases of asymptomatic reinfection only a very early or a very late serum was available. Of the remaining 19 cases, 7 had detectable specific IgG3. However, only one of 9 sera collected 30-50 days after contact contained specific IgG3. Thus for the asymptomatic patient for whom other serological tests suggest a recent rubella infection, the failure to detect specific IgG3 in sequential sera collected after contact suggests reinfection rather than primary rubella. The detection of specific IgG3 did not correlate with the presence of specific IgM. Sera collected 6-8 weeks after rubella vaccination had detectable specific IgG1 in 32 of 33 cases and specific IgG3 in 9 of 33. The remaining vaccinee was seronegative.