Systemic candidiasis, especially candidemia, is an increasing problem among high risk neonates. Although possible predisposing factors have been suggested, no case-control study has evaluated potential risk factors. By retrospective chart review we identified 21 infants admitted to the neonatal intensive care unit between 1976 and 1983 (0.9% of all admissions) who had documented nosocomial candidemia before 4 months of age. Twenty patients were matched to a control infant with similar birth weight and date of admission. We found that the median durations of exposure to the following risk factors were significantly longer in patients compared with controls: hyperalimentation; intravenous fat emulsion; endotracheal tubes; and/or tracheostomies and antibiotic therapy. However, by discriminant analysis, duration of antibiotic therapy remained the variable most strongly (and independently) associated with the development of candidemia. Treatment of the 22 episodes of candidemia was variable including catheter removal alone in 12 and a combination of catheter removal, amphotericin, 5-fluorocytosine and/or ketaconazole in the others. Although the overall mortality for the series was 5 of 20 cases (25%), no infant larger than 2000 g died. We conclude that development of candidemia in neonates is associated with, and possibly caused by, prolonged exposure to antibiotics (as documented by multivariate analysis), hyperalimentation, intravenous fat emulsion and tracheal intubation (as documented by univariate analysis). To the greatest extent possible consistent with good clinical care, exposure to these risk factors should be minimized in high risk neonates.