We retrospectively evaluated a possible difference in periods until start of insulin treatment between type 2 diabetic patients treated with gliclazide (GCZ) and glibenclamide (GBC), because GCZ might be protective for beta cells than GBC. Subjects were Japanese patients. GCZ group consisted of patients treated with GCZ alone or with GCZ and GBC in the separate treatment periods in combination with or without other oral hypoglycemic agents (OHAs), while GBC group consisted of patients with GBC alone or in combination with other OHAs except GCZ. The periods until the treatment of insulin commenced were calculated using the Kaplan-Meier method. Proportional hazards models were used to adjust the differing variables between GCZ and GBC groups. The periods until the start of insulin treatment from diabetes onset, diabetes treatment, or GCZ or GBC treatment were significantly longer in the GCZ group than those in GBC group (P<0.001 in each group). Independent variables affecting the period were average HbA1c levels during GCZ or GBC treatment (hazard ratio=2.5 per %), other OHAs combined (hazard ratio=1.9 on combination), and difference between GCZ and GBC groups (hazard ratio=0.5 on GCZ). These results imply that GCZ may be more protective against secondary beta cell failure than GBC.