23 subjects were followed prior to treatment for 5 to 12 months in an attempt to relate attachment loss during this period to attachment level, probing depth, gingival redness and bleeding on probing. The subjects were divided into 3 groups based on amount and distribution of prior attachment loss: minor periodontitis, predominantly molar periodontitis and generalized periodontitis. The % of sites that exhibited attachment loss during the study period in the minor periodontitis, predominantly molar periodontitis and generalized periodontitis groups, were 1.3, 8.1 and 5.4, respectively. Subjects with minor periodontitis and predominantly molar periodontitis exhibited attachment loss more frequently in molar sites, proximal sites and sites with attachment level greater than or equal to 4 mm. In subjects with generalized periodontitis, attachment loss was related to tooth surface and attachment level, but not to tooth type. The relationship between the clinical parameters and attachment loss was improved compared with previous studies by using more homogeneous subject groups, more sensitive methods of analyzing attachment change and multivariate data analysis. However, the clinical parameters could not be used as diagnostic tests to predict attachment loss at individual sites.