Abstract Standardized radiographs and repeated periodontal probe measurements were made on 22 untreated subjects with destructive periodontal disease monitored for I year. Radiographs of selected sites were taken at 0, 6 and 12 months. Measurements of attachment level were made monthly. Radiographic measurements were made on 7× magnified projected images. Alveolar bone height from the CEJ Was Computed by multiplying the average length of the root times the measured ratio of CEJ To Alveolar bone over CEJ To Root tip. Each radiograph was measured twice by 2 investigators. Sites were excluded as having indistinct anatomical landmarks in which the standard deviation of the 4 measurements exceeded 0.16 mm, the measurement error for repeat determination of bone height on high quality radiographic images. A 3 sigma critical value for significant bone loss was selected as 0.48 mm. Changes in attachment level were computed for the intervals preceding and during the 6–12 month radiographic measurement period. Based on these critical values, 6.1% of the 231 radiographed sites showed significant bone loss. Similarly, 5.7% of the 1155 probed sites showed significant attachment loss. However, none of the sites with significant bone loss exhibited significant attachment loss over the same time period. In general, significant attachment loss preceded bone loss by 6 to 8 months. At 4 mm, attachment loss was found to predict subsequent bone loss with a true positive ratio of 60% and a false positive ratio of 5%, indicating a high degree of predictive discrimination. These observations indicate that attachment loss precedes radiographic evidence of crestal alveolar bone loss during periods of periodontal disease activity.