Serum opsonization of yeasts for phagocytosis by normal polymorphonuclear leucocytes was defective in 11 of 43 children with unexplained frequent infections. The children had a range of infections, largely bacterial, and only 3 had diarrhoea and rash in infancy. A similar defect in at least 6 of the 9 mothers of these children (of either sex), with normal function in the fathers, suggests that the defect was primary and was transmitted by an unusual form of dominant inheritance. Four of 72 healthy adults and 1 of 11 children with unrelated disease showed similar defective function, but the incidence of the defect in the patients with frequent infection was significantly greater than this. The defective function can be corrected, in vitro and in vivo, by normal plasma at concentrations too low to be effective alone. This suggests that there is a defective factor rather than an inhibitor, and that different factors are limiting in normal and in defective plasma. Sera from affected members of the same family do not correct each other, but defective sera from different families usually do.