Serum and salivary responses of 95 infants to either a standard (4 x 10(4) plaque-forming units (PFU), 47 neonates) or a high dose (4 x 10(5) PFU, 48 neonates) of tetravalent reassortant rhesus rotavirus vaccine (administered at 2 days and at 6 weeks of age) were evaluated in a double-blind clinical trial. Serum and salivary IgA antibodies to the rotavirus group A common antigen were determined by ELISA and radioimmunoassay (RIA). Serum neutralizing antibodies to rhesus rotavirus were determined by fluorescent focus reduction assay. No significant differences in responses to the high versus standard dose were noted in serum or saliva. Response was influenced by cord blood antibodies. All infants who were cord blood-negative for rhesus rotavirus neutralizing antibodies (nine who received the standard dose and 20 who received the higher dose) had serum responses, compared with 42-70% of those who were cord blood-positive. The serum response rate recorded for babies with cord blood neutralizing titres > 1000 was 44%. Infants being bottle fed had a higher serum response rate than did babies being breast fed exclusively. If serum and salivary responses were combined, the response rate reached 80% for bottle fed infants. Thus, determination of serum responses alone underestimates vaccine 'take' in infants, and more so in highly endemic areas than in areas subject only to sporadic outbreaks. However, determination of salivary responses in newborn breastfed infants may be inaccurate, due to possible persistence of antibodies derived from colostrum or breast milk.