Forty children with atopic dermatitis were evaluated for clinical evidence of hypersensitivity to foods by double-blind, placebo-controlled food challenges. Twenty-four children (60%) experienced 33 positive challenges, manifested by cutaneous symptoms in 31 (94%), gastrointestinal symptoms in 14 (42%), nasal symptoms in nine (27%), and respiratory in six (18%). Results of prick skin tests (STs) and RASTs to eight food antigens frequently eliciting hypersensitivity reactions were compared with those from food challenges to determine the diagnostic accuracy in children with atopic dermatitis. Defining a positive ST as a wheal 3 mm larger than the negative control wheal and a positive RAST as a Phadebas RAST score of 3 or 4, the sensitivity, specificity, and predictive accuracies of these tests were found to be comparable except in the case of wheat antigen where the ST was clearly superior to the RAST. Accepting a RAST score of 2 or more as a positive slightly improved sensitivity in some cases but dramatically decreased specificity. Combining results of STs and RASTs did not improve significantly the diagnostic accuracy over results of the tests used individually. These studies demonstrate no advantage of RAST alone or in combination with prick skin testing over prick skin testing alone in the evaluation of food hypersensitivity in children with atopic dermatitis. Furthermore, skin testing should be considered a good test for excluding immediate food hypersensitivity but only a suggestive positive indicator of hypersensitivity due to the high rate of clinically insignificant positive STs.