Sixty six children with severe atopic eczema were treated with highly restricted ('few food') diets followed, if they improved, by serial reintroduction of excluded foods. Twenty four patients (36%) improved considerably during the few food phase of the diet. Fifteen of these (23% of the study group) maintained this improvement on dietary treatment, of whom three abandoned the diet after periods ranging from six to 10 months, despite continued benefit, because they found the dietary restrictions too arduous. Thus 12 out of 66 children (18%) with severe eczema experienced prolonged and useful benefit from this dietary manoeuvre. Double blind food challenges performed in 10 patients failed to establish that parental identification of provoking foods is reliable. A search for historical and in vitro predictors of diet responsiveness was unsuccessful in this series.