Background: Epidemiological surveys identify a 900 000-strong potential paediatric population who may self-report with food allergy in Italy. As the reference test is the double-blind, placebo-controlled food challenge for diagnosis, we carried out a nation-wide questionnaire survey of hospital-based paediatric allergy centres between 1 January and 30 June 2003. Methods: All registered centres received a 27-point questionnaire polling logistics and protocols of oral food challenge. Items polled included medical facilities and personnel, selection of patients, foods tested, blinding, placebo use, evaluation criteria, number of tests and positive outcome rates in 2002. Results: Two hundred and two of 456 responders (44.3%) carried out food challenges (a ratio of one centre to 37 400 children in the general population). 11 728 challenges were performed in 2002 (an average of 58 procedures per centre; range: 1–500) and 3350 procedures were claimed positive (28.6%). Cow's milk was tested by all centres, egg by 77.2%, beef by 19.8% and cereals, peanuts, fish or fruit collectively by 37.6%. Twenty-nine of 202 centres performed blinded tests, 18 of which under double-blinded conditions. In cow's milk challenges, placebos included lyophilized cow's milk capsules, soy formula, extensive whey hydrolysate or extensive casein hydrolysate. Conclusion: Our survey reveals great differences in challenge test criteria, facilities and procedures and that many centres perform relatively few procedures. This argues in favour of greater rationalization in the allocation of tertiary-level resources before procedure standardization can be contemplated in the diagnostic workup of food allergy.