Following active periodontal therapy, 41 patients were incorporated in a maintenance program for 2 1/2 years with recall intervals varying between 2-6 months. At the beginning of each maintenance visit, the periodontal tissues were evaluated using "bleeding on probing" (BOP). Reinstrumentation was only performed at sites which bled on probing. However, supragingival plaque and calculus was always removed. Pocket probing depths and probing attachment levels were recorded after active treatment and at the conclusion of the study. Progression of periodontal disease was defined by an observed loss of probing attachment of greater than or equal to 2 mm. The reliability of the BOP test as a predictor was evaluated by calculating sensitivity, specificity, accuracy, and positive and negative predictive values. While only a 29% sensitivity was calculated for frequent bleeding, the specificity was 88%. The fact that the positive predictive value for disease progression was only 6% and the negative predictive value was 98% renders continuous absence of BOP a reliable predictor for the maintenance of periodontal health.