In a previous report, it was shown that scaling and root planing (SRP) decreased mean pocket depth and attachment level in subjects with adult periodontitis, as well as the levels and prevalence of Bacteroides forsythus, Porphyromonas gingivalis and Treponema denticola. However, a subset of subjects in that study exhibited mean loss of attachment following SRP. The purpose of the present investigation was to seek clinical and microbiological differences between subjects who responded well or poorly to SRP. 57 subjects with adult periodontitis were treated by full-mouth SRP under local anaesthetic. Clinical assessments of plaque, redness, suppuration, BOP, pocket depth and attachment level were made at 6 sites per tooth prior to and 3 months post-SRP. Attachment level measurements were repeated at each visit and differences in means between visits used to assess change. 18 subjects showed mean attachment loss 3 months post-SRP (poor response group), while 39 showed mean attachment level gain (good response group). The prevalence and levels of 40 subgingival taxa in subgingival plaque samples from the mesiobuccal site of each tooth (maximum 28 sites) in each subject prior to and 3 months post-SRP were assessed using checker-board DNA-DNA hybridization. The prevalence of each species was computed for each subject and averaged across subjects in the 2 treatment-response groups at each visit. Differences between groups were sought using the Mann-Whitney test. There were no statistically significant differences between the 2 response groups in any clinical parameter prior to therapy. Subjects in the good response group showed more attachment level gain at sites with baseline pocket depths of < 4 mm, 4-6 and > 6 mm than poor response subjects. Of 40 species evaluated, A. naeslundii genospecies 2 (A. viscosus), T. denticola, C. gracilis and C. rectus were significantly higher and more prevalent pre-therapy in the good response subjects. Mean attachment level change post SRP could be predicted using multiple linear regression with A. naeslundii genospecies 2 (A. viscosus) and T. denticola as the predictor variables (r2 = 0.373, p < 0.00001). Sites that gained > or = 2 mm of attachment post therapy showed a significant decrease in the counts of P. gingivalis (7.5 +/- 3.5 to 0.2 +/- 0.2 x 10(5)), T. denticola (8.2 +/- 3.5 to 1.8 +/- 1.1 x 10(5)) and B. forsythus (11.1 +/- 5.7 to 0.3 +/- 0.2 x 10(5)). The data of the present investigation indicate that SRP is most effective in subjects and sites with high levels of the subgingival species that this therapy affects.