Panoramic, bitewing and periapical radiography and probing for measurement of the marginal bone level were compared. Altogether 237 sites of 23 patients were examined. Radiographs were taken with a splint containing steel balls to allow calculation of the enlargement of the radiographs. Probing was done before and during flap surgery using the same splint. The open bone measurement represented the true value. All radiographs were assessed by 5 observers. The mean enlargement of panoramic radiography was 27% in the upper and 26% in the lower arch. For bitewing and periapical radiography, it was 8% in the upper and 4-5% in the lower arch. All methods underestimated the bone loss. Probing bone level before surgery was most accurate, deviating at most 5% from the true value. Periapical radiography was more accurate than panoramic and bitewing radiography (p less than 0.001). Panoramic radiography presented a slightly lower mean accuracy than bitewing radiography (p less than 0.05). The underestimation of the bone loss ranged from 13 to 32% in orthopantomograms, 11-23% in bitewing and 9-20% in periapical radiographs. The interobserver variation of the radiographic methods was substantial.