This study aimed to explore the effects of pioglitazone treatment on progression from persistent atrial fibrillation (AF) to permanent atrial fibrillation in diabetes mellitus (DM) patients and to investigate the possible mechanisms involved in those effects. A total of 146 diabetes mellitus (DM) patients with firstly identified persistent AF were selected. Seventy patients were randomized into the pioglitazone (30 mg/day) group and 76 into the placebo group. Pro-collagen type I carboxyterminal peptide (PICP), advanced glycation end products (AGEs), and angiotensin II were assayed and left atrial diameter (LA diameter) was measured at the first presence of persistent AF, and at 6 and 14 months of follow-up. The time point of identification of permanent AF and the incidence of permanent AF in the patients were all recorded. Thirty-seven (49%) of the 76 patients in the placebo group and 21(30%) of the 70 patients in the pioglitazone group progressed to permanent AF (P = 0.028). No significant differences existed in the follow-up time (20.5 +/- 3.97 months for pioglitazone group versus 20.9 +/- 4.14 months for placebo group) between the two groups (P = 0.535). In the pioglitazone group, no significant change was found in angiotensin H level. The PICP level did not change significantly at 6-months of follow-up, but decreased significantly at 14-months of follow-up (P = 0.032). The AGE (P = 0.037 at 6-month follow-up, P < 0.035 at 14-month follow-up) level was significantly lower at both 6 and 14-months of follow-up. By lowering the PICP level, pioglitazone treatment may decrease the incidence of permanent AF in DM patients with persistent AF, which may be associated with the suppressing effect of pioglitazone on AGEs.