The relationship between the intestinal colonization of a test strain of Bifidobacterium bifidum requiring human milk growth-promoting factors in vitro and the presence of growth-promoting factors either in the stools of human neonates or in their diet was investigated. Thirty-one infants were inoculated with a single dose of this strain within the first 8 d of life. Spores of a strictly thermophilic Bacillus admixed with the B. bifidum inoculum were used as transit marker, and the fecal population levels of both strain B. bifidum and the transit marker were recorded within 6 d after inoculation. Strain B. bifidum was found in the predominant flora of six neonates. It was eliminated more quickly than the transit marker from the stools of 17 neonates. Its population remained at a low level in the remaining eight neonates. Amounts of B. bifidum growth-promoting factor in the infant stools were not significantly different whether they harbored strain B. bifidum at a high population level or not. Although these amounts were significantly higher in infants fed human milk containing B. bifidum growth-promoting factor than in infants fed formula milk without B. bifidum growth-promoting factor, strain B. bifidum became established in one of the 18 infants fed human milk and in five of the 13 formula-fed infants. No relationship could be found between the population levels of strain B. bifidum and those of facultatively anaerobic streptococci and enterobacteria already present on d 0 and 1. These results clearly show that no relationship exists between the intestinal colonization of B. bifidum and the amounts of exogenous or endogenous growth-promoting factors found in stools.