To determine whether simple protective isolation reduces the incidence of nosocomial bacterial and fungal infection during pediatric intensive care, we randomly assigned 70 children who were not immuno-suppressed and who required mechanical ventilatory support and three or more days of intensive care to receive standard care (n = 38) or protective isolation (n = 32) with use of disposable, non-waven, polypropylene gowns and nonsterile latex gloves. Risk factors predisposing patients to infection were comparable in the two groups. Nosocomial colonization occurred later among isolated patients (median, vs. 7 days; P less than 0.01) and was associated with subsequent infection in 12 patients, as compared with 12 patients given standard care (P = 0.01). Among patients who were isolated, the interval before the first infection was significantly longer than (median, 20 vs. 8 days; P = 0.04), the daily infection rate was 2.2 times lower than (95 percent confidence interval, 1.2 to 4.0; P = 0.007), and there were fewer days with fewer (13 percent vs. 21 percent; P = 0.001). The benefit of isolation was most notable after seven days of intensive care. Isolation was well tolerated by patients and their families. Regular monitoring showed that the children in each group were touched and handled comparably often by hospital personnel and family members. We conclude that the use of disposable, high-barrier gowns and gloves for the care of selected, high-risk children who require prolonged intensive care significantly reduces the incidence of nosocomial infection, is well tolerated, and does not compromise the delivery of care.