This paper reports the results of a longitudinal study of Canadians aged 50 years and over which was designed to estimate the three-year incidence of root caries. At baseline, 699 randomly selected dentate subjects were clinically examined and 493 were examined again after three years. Over this period 27.4% had one or more root DFS increments and the mean DFS increment was 0.6 per person. Because DFS increments overstate root caries incidence DS increments were also calculated. Relatively few baseline variables were associated with either DFS or DS increments. In logistic regression analyses, age was the only variable associated with one or more root DFS increments, while age, dental visiting pattern and wearing a partial denture were associated with one or more root DS increments. In both cases, the predictive power of the models was poor but improved marginally when baseline root caries experience was also entered as an independent variable.