Selective rubella vaccination of 12-year-old schoolgirls was introduced in Sweden in 1973 and at the same time a long-term follow-up cohort study was initiated. In 1982, a two-dose programme with a combined vaccine against measles, mumps and rubella (MMR) was introduced and vaccinations were given at the ages of 18 months and 12 years to both boys and girls. The cohort initially comprised 486 girls. It was followed for between 8 and 16 years. All the girls enrolled were seronegative before vaccination and had seroconverted to a haemagglutination-inhibition (HAI) titre of at least 1:16. On the last test occasion 16 years later, 22% had titre values below 1:16, and 6% lacked detectable antibodies against rubella (< 1:8). A fourfold or greater rise in titre was seen in 36% of the girls during the first 8 years of observation, whereas during the following 8 years only 1% showed a significant increase of titre values. The geometric mean titre declined from 1:110 to 1:34 during the first 8 years and further to 1:18 during the following 8 years. From 1982 to 1990, the seroimmunity to rubella of 18-year-old girls and boys was studied yearly. The number studied was 3308 18-year-old schoolgirls and 6347 18-year-old recruits born between 1964 and 1972. The recruits were divided into two groups, 4610 unvaccinated and born in 1964-1969 and 1737 vaccinated and born in 1970-1972. Seropositive recruits in the first group were thus naturally immune only, while the second group had a mixture of natural and vaccine-induced immunity.(ABSTRACT TRUNCATED AT 250 WORDS)