During a community-wide outbreak of hepatitis A in Gloucester, UK there was a high attack rate in children attending two city primary schools and a pre-school centre sharing the same site. In September 1990, saliva specimens were collected from 478 (85%) of the 562 children. The prevalence of antibody to hepatitis A virus (anti-HAV), as determined by saliva testing, was 29.6%; highest prevalences were seen in 5-6-year-olds and in children from that area of the city at the centre of the community-wide outbreak. The proportion of immune children with a history of clinical hepatitis varied with age from 1 in 42.7 of under-5-year-olds to 1 in 4.7 of 8-10-year-olds. Six children who received prophylaxis with human normal immune globulin (HNIG) because they were household contacts of cases subsequently became infected. Since there was evidence of transmission outside the school environment it is unlikely that a policy of universal prophylaxis within the schools would have stopped the outbreak. Mass prophylaxis in school outbreaks is only likely to be effective if most transmission is occurring at school and if the target population can be clearly defined. Salivary antibody testing is a simple, practical and acceptable procedure in young children. Salivary antibody surveys in conjunction with vaccination against hepatitis A should provide a cost-effective method for control of future outbreaks.