An observer strategy involving trained and experienced examiners was used in a large series of endodontic surgery cases, evaluated radiographically 1 year after the operation. The cases were grouped in 1 of the following 4 healing groups; complete healing, incomplete healing (scar tissue), uncertain healing and unsatisfactory healing (failures). The radiographs were first examined separately by an oral surgeon and an endodontist. Deviating cases were subjected to joint evaluation. The overall agreement before joint interpretation was satisfactory: 63% when corrected for chance agreement. Agreement after joint evaluation reached 94% of the total material. Disagreement and difficult borderline cases were judged by an oral radiologist. He was more pessimistic than the 2 main examiners. The identification of healed cases with apical bone which had not yet obtained normal radiopacity and structure was difficult. Altogether, the observer consistency was satisfactory; the selection of experts increased the chances of "true" diagnoses; the important screening function of the one-year control was confirmed. The applied strategy and the classification are recommended for follow-up studies.