Background: Exposure to specific bacterial bowel commensals may increase/reduce the risk of atopic diseases. Objective: To compare fecal bacterial communities of young infants with/without eczema. Methods: Nested case-control study. Infants age 3 to 6 months with eczema (cases, n = 37) and without (controls, n = 24) were matched for sex, age, feeding (breast/bottle/mixed/solids), ethnicity. Information was collected on maternal/infant antibiotic exposure, feeding, gastrointestinal symptoms, family history of allergy. Eczema severity scoring was used (Severity Scoring of Atopic Dermatitis index). Samples were taken for determination of allergen-specific serum IgE (cases) and urinary/fecal eosinophilic protein X. Gastrointestinal permeability was measured. The compositions of fecal bacterial communities were analyzed (culture-independent, nucleic acid-based analyses). Results: There was no difference in overall profiles of fecal bacterial communities between cases and controls. Family history of allergy increased likelihood of bifidobacteria detection (history, 86%; no history, 56%; P = .047); breast-fed infants were more likely to harbor Bifidobacterium bifidum (odds ratio, 5.19; 95% CI, 1.47-18.36; P = .01). Bifidobacterium pseudocatenulatum was detected more commonly in feces of non-breast-fed children (odds ratio, 5.6; 95% CI, 1.3-24.3; P = .02) and children with eczema (eczema, 26%; no eczema, 4%; P = .04). There were no significant associations between clinical measurements and detection of B pseudocatenulatum. Conclusion: Presence of B pseudocatenulatum in feces was associated with eczema and with exclusive formula-feeding; B bifidum was associated with breast-feeding. © 2008 American Academy of Allergy, Asthma & Immunology.