The periodontal pathogen Actinobacillus actinomycetemcomitans can frequently be isolated from subgingival plaque of adults with chronic inflammatory periodontal disease and individuals with plaque-induced gingivitis. Problems with the persistence of the organism after thorough debridement of root surfaces have been reported. In the present study clinical effects of the hygienic phase of periodontal therapy in ten adult patients with moderate or advanced periodontitis harbouring A. actinomycetemcomitans were analysed. Since proper analysis of highly correlated data within a given patient is crucial for appropriate interpretation, a major objective of this study was to compare the results of different models derived from logistic regression of clinical and microbiological factors on gain or loss of clinical attachment under different assumptions. Subgingival samples from every tooth present were obtained before and 6 weeks after thorough subgingival scaling, and selectively cultivated for the organism. A relevant gain of clinical attachment of 2 mm or more was observed at a total of 36% of periodontitis sites after scaling. Overall, loss of attachment of 2 mm or more was observed at 8% sites. Most loss occurred at sites with gingival enlargement (15%), whereas 3% periodontitis sites lost 2 mm or more. In multivariate analyses erroneously assuming either independence of data or correctly considering the correlated structure of observations attachment gain was mainly associated with deep probing depths at the outset. Presence or absence of A. actinomycetemcomitans before or after therapy was not included into the periodontitis models. Also, loss of attachment of 2 mm or more after subgingival scaling was not influenced by the organism. A direct comparison of the results obtained with both approaches of logistic regression may be helpful in the assessment of the influence of the magnitude of correlation of the data on the regression coefficients.