Background Mononuclear cells from children with active atopic dermatitis (AD) have been reported to be hyper-responsive to certain microbial stimuli, in particular staphylococcal enterotoxin B (SEB). However, it is not known whether this responsiveness is acquired during disease development, or is inherent. We investigated this question in a cohort of children at high risk of atopy followed prospectively from birth to age 3 years. We asked whether their cord blood mononuclear cell (CBMC) cytokine responses to SEB, to an unrelated microbial stimulus purified protein derivative (PPD), or to common allergens, were predictive of risk for subsequent AD development during infancy.Methods Children at high risk of developing atopy were randomly selected from an ongoing prospective cohort. Cord blood was collected at birth. The children were seen at 6 months, 1, 2 and 3 years and examined for the development of AD. IFN-γ, IL-5, IL-10 and IL-13 production by CBMC cultured in the presence of SEB, PPD, PHA, house dust mite (HDM) allergen, ovalbumin (OVA) and cat allergen was determined.Results SEB-induced IL-5 production by CBMC was elevated in children who developed AD at 6 months (P = 0.01) and 2 years (P = 0.009). PPD-induced IL-5 responses were also elevated in CBMC from children who developed AD at 6 months, 2 years and 3 years (P = 0.05, P = 0.06 and P = 0.06, respectively), as were PPD-induced IL-10 responses (P = 0.05 at 1 years, P = 0.007 at 2 years, P = 0.003 at 3 years) and corresponding IFN-γ responses (P = 0.05 at 6 months, P = 0.003 at 2 years, P = 0.0004 at 3 years). Increased IL-10 responses to HDM allergen were also observed throughout the observation period in CBMC from children who developed AD.Conclusion Children who develop infantile AD appear to have a predisposition to respond to SEB in a Th2-dominant manner involving selective stimulation of IL-5 production. The increased IL-10 and IFN-γ induced in response to PPD by children with AD may point to additional intrinsic differences in responses to microbial stimuli between those at high vs. those at low risk for AD, which merit more detailed investigations.