The most common forms of destructive periodontal disease have been thought to slowly and continuously progress until treatment or tooth loss. Recently, data have become available which are inconsistent with this "continuous disease" hypothesis. Data from longitudinal monitoring of periodontal attachment levels and alveolar bone in humans and in animals suggest that periodontal disease progresses by recurrent acute episodes. In addition, rates of attachment loss have been measured in individual sites which are faster than those consistent with the continuous disease hypothesis or slower than those expected from estimates of prior loss rates. To account for these observations, a model of destructive periodontal disease is described in which bursts of activity occur for short periods of time in individual sites. These bursts appear to occur randomly at periodontal sites throughout the mouth. Some sites demonstrate a brief active burst of destructive periodontal disease (which could take a few days to a few months) before going into a period of remission. Other sites appear to be free of destructive periodontal disease throughout the individual's life. The sites which demonstrate destructive periodontal activity may show no further activity or could be subject to one or more bursts of activity at later time periods. Comparison of monitored loss rates for a year with mean loss rates prior to monitoring suggested that there may be relatively short periods in an individual's life in which many sites undergo periodontal destruction followed by periods of extended remission. An extension of the random disease model is also suggested in which bursts of destructive periodontal disease activity occur with higher frequency during certain periods of an individual's life.