In a joint prospective study in Germany and the United Kingdom between 1980 and 1993, 1373 women who had varicella and 366 who had herpes zoster during the first 36 weeks of gestation were followed up. 9 cases of congenital varicella syndrome were identified, all occurring after maternal varicella during the first 20 weeks of gestation. The highest risk (2.0%) was observed between 13-20 weeks gestation, with 7 affected infants identified among 351 pregnancies (95% CI of risk 0.8-4.1%). Only 2 cases of congenital varicella syndrome were identified among 472 pregnancies in which maternal varicella occurred before 13 weeks (observed risk 0.4%, 95% CI 0.05-1.5%). Herpes zoster in infancy was reported in 10 children whose mothers had had varicella in pregnancy. No infants with clinical evidence of intrauterine infection were born to the 366 women with herpes zoster in pregnancy (upper 95% confidence limit of estimated risk 1.0%). Varicella-zoster-specific IgM antibody was found at birth in 4 of 16 (25%) infants with clinical manifestations of intrauterine infection and persistent specific IgG antibody in 5 of 7 infants tested. The corresponding rates in asymptomatic infants whose mothers had varicella were 12% (76/615) and 7% (22/335) respectively. No serological evidence of intrauterine infection was found in infants who mothers had herpes zoster in pregnancy. In 97 pregnant women, varicella occurred after post-exposure prophylaxis with anti-varicella-zoster immunoglobulin. No cases of congenital varicella syndrome or zoster in infancy occurred in this group. Our estimates provide a sound basis for counselling women with varicella in pregnancy. Although the risk of congenital varicella syndrome is small, the outcome for the affected infant is so serious that a reliable method of prenatal diagnosis would be valuable. In the long term, prevention of maternal varicella would be an option if a safe and effective vaccine were to become routinely available.