Microbiologic assessments are often included in longitudinal studies to elucidate the significance of the association of certain Gram-negative bacteria and the development of periodontal diseases. In such studies, the reliability of methods is crucial. There are several methods to identify putative pathogens, and some of them are commercially available. The purpose of the present study was to compare the reproducibility of four different methods for detecting Actinobacillus actinomycetemcomitans, Porphyromonas gingivalis, and Prevotella intermedia in order to evaluate their usefulness in epidemiologic studies. The test panel consisted of 10 young subjects and 10 adult periodontitis patients. Subgingival plaque was sampled from sites showing bone loss and "healthy" control sites. The four different methods for detecting the target bacteria were 1) cultivation, 2) Evalusite (a chair-side kit based on ELISA), 3) OmniGene, Inc, based on DNA probes, and 4) indirect immunofluorescence (IIF). The test procedure was repeated after a 1-wk interval and was performed by one examiner. Sites reported to be positive for a microorganism by any of the four methods at one or both examinations were considered to be positive for that organism and included in the analysis. The reproducibility of the four methods was low. The IIF and the cultivation methods showed somewhat higher reproducibility than did the commercial systems. A second test was done for Evalusite, three paper points for sampling being used instead of one as described in the manual. The reproducibility of the second test was improved, indicating that the detection level of the system may influence the reliability.