We developed a computerized record linkage system to determine whether it is feasible to use self-administered questionnaires to conduct large-scale, routine postdischarge surveillance for nosocomial infections among mothers and infants. We mailed questionnaires to 19,650 mothers of infants delivered at our institution who were discharged between January 1, 1988, and December 31, 1989. The questionnaire asked whether either mother or baby experienced infections, received an antibiotic, or was rehospitalized for an infection. Thirty-six percent of mothers returned the questionnaire. Seven hundred eleven (10%) of 7,033 replies reported 763 infections in mothers and 844 (12%) of 7,147 replies reported 968 infections in infants. These infections represented 4% of all women who delivered and 4% of all infants discharged. Since a minority of questionnaires were returned, these are likely to be conservative estimates of the occurrence of perceived infections. The most common maternal infections were mastitis (406 cases), urinary tract infections (185 cases), and endometritis (58 cases). The most common infections of infants were conjunctivitis (622 cases), diaper rash requiring antibiotics (169 cases), and umbilical infection (84 cases). Eighty-two percent of maternal infections reported after discharge were treated with antibiotics, as were 74% of infant infections. Infections reported during a 4-mont period were checked by a telephone call to the mother, who confirmed 17 (74%) of 23 reported maternal infections and 40 (70%) of 57 reported infant infections. Infections reported during a separate 3-month period were checked by contacting the patients' physicians, who confirmed 23 (48%) of 48 reported maternal infections and 19 (25%) of 76 reported infant infections. It is unclear how much of the low proportion of reported cases confirmed by physicians reflects overreporting by mothers and how much represents incomplete knowledge by physicians. In addition, some reported infections may have been acquired after discharge. Postdischarge surveillance via self-administered questionnaire identified twice as many apparent maternal infections and 12 times as many newborn infections as did concurrent prospective in-hospital surveillance. However, a number of important methodologic issues must be resolved.