The efficacy of metronidazole and doxycycline in preventing recurrent periodontitis was studied in 23 patients. After treatment in the previous 7 months with either bimonthly scaling and 3 weeks of systemic doxycycline (11 subjects) or scaling and placebo (12 subjects), patients were monitored for recurrent periodontitis and were scaled every 2 months. When either a periodontal abscess or greater than 2 mm loss of gingival attachment was observed, metronidazole was administered (250 mg every 8 hours) for 10 days. In the placebo plus metronidazole group, 5 patients (42%) exhibited recurrent periodontitis after the metronidazole regimen compared with only one (9%) in the doxycycline plus metronidazole group (P less than 0.096). Subgingival plaque samples at study and healthy control sites were screened for the presence of Actinobacillus actinomycetemcomitans, Porphyromonas gingivalis, Prevotella intermedia, Eikenella corrodens, and Fusobacterium nucleatum by immunofluorescence and for spirochetes using Ryu's stain. Presence/absence analysis of the sum of scores of the 6 individual pathogens demonstrated large reductions (P less than 0.005) in the frequency of detection of pathogens in the former doxycycline compared with the placebo plus metronidazole group at both study and control sites before and one month after metronidazole. By 7 months after metronidazole, there was no detectable difference between groups. These results indicate that prevention of recurrent periodontitis with metronidazole may be enhanced by previous treatment with doxycycline.