OBJECTIVE Optimal insulin regimens for non-insulin-dependent diabetes mellitus (NIDDM) patients with secondary failure are controversial. We evaluated the efficacy, side effects, and quality of life of patients receiving insulin either alone or in combination with their previous oral hypoglycemic agents (OHAs). METHODS Fifty-three Chinese patients with NIDDM (mean age 53.9 +/- 12.6 years, duration of diabetes 9.0 +/- 4.9 years, body wt 60.4 +/- 13.3 kg with corresponding body mass index 24.2 +/- 4.3 kg/m2, receiving the maximum dose of sulfonylurea and/or metformin) were confirmed to have OHA failure. Twenty-seven patients were randomized to continue OHAs and were given additional bedtime insulin (combination group); 26 patients were randomized to insulin therapy alone with twice-daily insulin (insulin group). Insulin doses were increased incrementally, aiming at fasting plasma glucose (FPG) < 7.8 mmol/l during a stabilization period of up to 8 weeks. Insulin dosage, body weight, glycemic control, and quality of life were assessed before and at 3 and 6 months after stabilization. RESULTS Both groups showed similar improvement of glycemic control. For the combination group, FPG decreased from 13.5 +/- 2.7 to 8.9 +/- 3.0 mmol/l at 3 months (P < 0.0001) and to 8.6 +/- 2.5 mmol/l at 6 months (P < 0.0001). For the insulin group, FPG decreased from 13.5 +/- 3.6 to 7.5 +/-3.0 mmol/l at 3 months (P < 0.0001) and to 9.8 +/- 3.5 mmol/l at 6 months (P < 0.0001). No significant differences were observed between the groups. Similarly, both groups had significant improvement of fructosamine and glycosylated hemoglobin (HbA1c). Fructosamine fell from a mean of 458 to 365 mumol/l at 3 months (P < 0.0001) and to 371 mumol/l at 6 months (P < 0.0001) and from 484 to 325 mumol/l at 3 months (P < 0.0001) and to 350 mumol/l at 6 months (P < 0.0001) for the combination and insulin groups, respectively. HbA1c decreased from 10.2 to 8.4% at 3 months (P < 0.0001) and to 8.7% at 6 months (P < 0.0001) in the combination group and from 10.7 to 7.8% at 3 months (P < 0.0001) and to 8.4% at 6 months (P < 0.0001) in the insulin group. Despite similar improvement of glycemia, insulin requirements were very different. At 3 months, the combination group was receiving a mean of 14.4 U/day compared with 57.5 U/day in the insulin group (P < 0.0001). Similar findings were observed at 6 months (15.0 vs 57.2 U/day, P < 0>0001). Both groups gained weight. However, for the combination group, weight gain was 1.6 +/- 1.8 kg at 3 months and 2.1 +/- 2.5% kg at 6 months (both P < 0.0001 vs baseline), whereas for the insulin group, weight gain was 3.5 +/- 4.3 and 5.2 +/- 4.1 kg, respectively (both P < 0.0001 vs baseline). Weight gain was significantly greater in the insulin group (P < 0.05 at 3 months, and P < 0.005 at 6 months). Fasting plasma triglyceride decreased in the insulin group (1.8 +/- 1.0 to 1.4 +/- 0.8 mmol/l at 3 months [P < 0.005] and to 1.4 +/ 0.7 mmol/l at 6 months [P < 0.02] but not in the combination group. No changes were observed in total and high-density lipoprotein cholesterol. No severe hypoglycemic reactions were recorded in either group. Mild reactions occurred with similar frequency in both groups. Well-being and quality of life improved significantly in both groups. The majority of patients (82.7%) wanted to continue insulin beyond 6 months, irrespective of the treatment group. CONCLUSIONS In NIDDM patients with secondary OHA failure, therapy with a combination of OHAs and insulin and with insulin alone was equally effective and well tolerated. However, combination therapy was associated with a lower insulin dose and less weight gain. Combination treatment may be considered when OHA failure occurs as a potential intermediate stage before full insulin replacement.