In two trials of measles vaccination in Guinea-Bissau, children were randomized to receive either the Edmonston-Zagreb (EZ) virus at age 4-8 months or, as a control group, a standard dose (5000 p.f.u.) of the Schwarz (SW) virus at 9-12 months. In the first trial a medium dose of EZ virus (40,000 p.f.u.) was used and in the later trial a high dose (150,000 p.f.u.). Pre- and postvaccination blood samples were analysed with a haemagglutination inhibition (HAI) test and an enzyme-linked immunosorbent assay (ELISA). For the ELISA the blood samples were collected on filter paper. Of 362 prevaccination specimens, 18 and 55%, respectively, were positive in the ELISA and HAI tests. At 18-20 months, the children who had received the EZ vaccine had a seropositivity of 91% by the ELISA method and one of 89% by the HAI method. The equivalent values for the SW group of children were 100 and 96% respectively. Antibody levels in the EZ group, as measured by either method, were significantly lower than the levels in the SW group. The serological results of the present study suggest that lowering the age at measles vaccination to below 9 months is feasible. However, further studies are needed to determine which virus strain, dosage and age at vaccination will prove to be optimal in countries where severe measles is common before the age of 9 months.