This study tested the hypothesis that the risk for alveolar bone loss is greater, and bone loss progression more severe, for subjects with poorly controlled (PC) type 2 diabetes mellitus (type 2 DM) compared to those without type 2 DM or with better controlled (BC) type 2 DM. The PC group had glycosylated hemoglobin (HbA1) > or = 9%; the BC group had HbA1 < 9%. Data from the longitudinal study of the oral health of residents of the Gila River Indian Community were analyzed. Of the 359 subjects, aged 15 to 57 with less than 25% radiographic bone loss at baseline, 338 did not have type 2 DM, 14 were BC, and 7 were PC. Panoramic radiographs were used to assess interproximal bone level. Bone scores (scale 0-4) corresponding to bone loss of 0%, 1% to 24%, 25% to 49%, 50% to 74%, or > or = 75% were used to identify the worst bone score (WBS) in the dentition. Change in worst bone score at follow-up, the outcome, was specified on a 4-category ordinal scale as no change, or a 1-, 2-, 3-, or 4-category increase over baseline WBS (WBS1). Poorly controlled diabetes, age, calculus, time to follow-up examination, and WBS1 were statistically significant explanatory variables in ordinal logistic regression models. Poorly controlled type 2 DM was positively associated with greater risk for a change in bone score (compared to subjects without type 2 DM) when the covariates were included in the model. The cumulative odds ratio (COR) at each threshold of the ordered response was 11.4 (95% CI = 2.5, 53.3). When contrasted with subjects with BC type 2 DM, the COR for those in the PC group was 5.3 (95% CI = 0.8, 53.3). The COR for subjects with BC type 2 DM was 2.2 (95% CI = 0.7, 6.5), when contrasted to those without type 2 DM. These results suggest that poorer glycemic control leads to both an increased risk for alveolar bone loss and more severe progression over those without type 2 DM, and that there may be a gradient, with the risk for bone loss progression for those with better controlled type 2 DM intermediate to the other 2 groups.