OBJECTIVE Little is known about late phase clinical reactions during oral food challenges and the value of specific IgE in terms of sensitivity and specificity. METHODS We therefore analysed retrospectively the clinical outcome of 387 oral provocations during double-blind, placebo-controlled food challenge tests in 107 children with atopic dermatitis. RESULTS Eighty-seven (81%) children showed a positive clinical reaction to at least one challenge. The vast majority of children (94%) showed clinical symptoms to one or two allergens. One hundred and thirty-one of 259 (51%) of verum challenges and 1/128 (0.8%) placebo challenge were assessed as positive. Oral provocations with hen's egg showed the highest percentage of positive reactions (70%). Sensitivity of specific IgE to the four allergens tested was 90%, specificity 30%. Sensitivity of the parental history as a predictive factor was 48%, specificity 72%. Ninety-two of 131 (70%) children with positive verum provocations showed early reactions, 33 (25%) late and six (5%) combined early and late reactions. In 84/131 (64%) positive provocations one organ system was involved, while in 44 (34%) provocations two and in three (2%) challenges three organ systems were involved. Skin reactions were the most frequent clinical manifestation leading to positive reactions followed by gastro-intestinal and respiratory symptoms. There was no correlation between titration dose and specific IgE. The subgroup of non-sensitized children did not differ in terms of sex, age or titration dose from the total study population. CONCLUSIONS Double-blind, placebo-controlled oral food challenges are helpful in distinguishing children with clinically manifested symptoms from those with food sensitization. Accurately identifying children with a clinical relevant food allergy may help to prescribe specific diets on a scientific basis, avoiding dietary limitations which may be unnecessary or even harmful.