Two-way relationship between periodontitis and diabetes was advocated; however, bidirectional observation in general population is still inconclusive. Using the Taiwan Health Insurance Database (covering over 99% of the entire population),11,011 patients with severe periodontitis were recruited from 2000 to 2015.After matching by age, sex, and index date, 11,011 patients with mild periodontitis and 11,011 non-periodontitis controls were registered. The outcome of T2DM was traced. Conversely, the development of periodontitis was traced in 157,798 patients with T2DM, and 157,798 non-diabetic controls enrolled. The risks of T2DM significantly increased in groups with severe and mild periodontitis, with the adjusted hazard ratio (aHR) and 95% confidence interval (CI) being 1.94 (1.49–2.63, p < 0.01) and 1.72 (1.24–2.52, p < 0.01), respectively. Patients with severe periodontitis had a high risk of having diabetes compared to those with mild periodontitis [aHR, 1.17 (95% CI 1.04–1.26, p < 0.001)]. Conversely, the risk of periodontitis increased significantly in patients with T2DM [1.99 (1.42–2.48, p < 0.01)]. However, the high risk was not observed for the outcome of mild periodontitis [0.97 (0.38–1.57, p = 0.462)]. We, therefore, suggested the bi-direction is between diabetes and severe periodontitis, but not in mild type.