BACKGROUND: Early intensive insulin therapies in newly diagnosed type 2 diabetic patients may improve beta-cell function and yield prolonged glycemic remissions. This study was performed to evaluate the relationship between the glycemic remission and beta-cell function and assess the variables predictive of long-term near-normoglycemic remission. METHODS: Eighty-four newly diagnosed type 2 diabetic patients were treated with 2-week continuous subcutaneous insulin infusion (CSII) and followed up longitudinally. Intravenous glucose tolerance tests (IVGTTs) were performed, and blood glucose, hemoglobin A1c (HbA1c) and insulin were measured at baseline, after CSII and at 2-year visit. The patients who maintained glycemic control for two years were defined as the remission group and those who relapsed before the 2-year visit were the non-remission group. RESULTS: The duration to be diagnosed of the patients (from the time that patients began to have diabetic symptoms until diagnosis) in the remission group was shorter than that in the non-remission group (1.00 month vs 4.38 months, P = 0.040). The increase of the acute insulin response (AIR) was maintained after 2 years in the remission group compared with AIR measured immediately after intervention (413.05 pmol*L(-1)*min(-1) vs 408.99 pmol*L(-1)*min(-1), P = 0.820). While AIR in the non-remission group significantly declined (74.71 pmol*L(-1)*min(-1) vs 335.64 pmol*L(-1)*min(-1), P = 0.030). Cox model showed that a shorter duration to be diagnosed positively affected the duration of near-nomoglycemic remission with an odds ratio (OR) 1.019, P = 0.038, while fasting plasma glucose (FPG) and post-breakfast plasma glucose (PPG) after CSII were the risk factors (OR 1.397, P = 0.024 and OR 1.187, P = 0.035, respectively). CONCLUSION: The near-normoglycemic remission is closely associated with long-term maintenance of beta-cell function and occurs more commonly in patients with shorter duration to be diagnosed and better glycemic control during CSII.