The aim of this retrospective study was to determine if a chairside assay for neutral protease activity in gingival crevicular fluid (GCF) could provide an early indication of site-specific disease activity, as defined by probing attachment loss. Clinical measures and assay data were collected at 6-month intervals over an average period of 26 months (range from 24 to 36 months). 38 subjects were selected from a pool of previously-treated chronic adult periodontitis patients who were in periodontal maintenance at intervals of 3 to 6 months. GCF samples were collected from a total of 71 tooth sites exhibiting prior attachment loss, and analyzed for neutral protease activity (NPA). Positive BOP and positive NPA scores were classified as true positives if sites subsequently lost at least 1 mm of attachment over the next 12 months, as confirmed by linear regression analysis. As a predictor of breakdown, the NPA assay had an accuracy of 94% and a risk ratio of 37.6 as compared to values of 58% and 1.5 for BOP. When only the subset of sites > or = 7 mm were considered, the NPA assay had a calculated accuracy of 92% versus a value of 50% for BOP. These results indicate that the assay appears to differentiate between bleeding at sites exhibiting only chronic inflammation with no attachment loss and bleeding at sites undergoing active attachment loss.