Secondary failure and the requirement is common in patients with non-insulin dependent diabetes mellitus. The combination of sulfonylureas with NPH insulin at bedtime has been proposed to avoid high doses of insulin. We treated 18 patients (2 men, age range 47-76 yr) non respondent to diet and glibenclamide, combining NPH insulin in an average dose of 0.3 +/- 0.03 U/kg BW at bedtime for 6 months. Fasting serum glucose improved from 256 +/- 11 to 132 +/- 6 mg/dl and HbA1C from 13.6 +/- 0.4 to 9.9 +/- 0.2%. Four patients achieved a good control (defined as a HbA1C < 9), 9 a fair control (HbA1C 9.1-10) and 5 persisted with a bad control (HbA1C > 10). Well controlled patients were younger, had a shorter duration of diabetes and had a non significantly higher body mass index. Fasting serum insulin and C peptide levels achieved after glucagon injection were not predictors of the metabolic response to combined therapy. Tolerance to treatment was good, without changes in blood pressure or serum lipids and with a low incidence of hypoglycemia. There was a mean increase of 3.6 kg in body weight. After 6 months of therapy, maximum achieved C peptide values after glucagon increased from 3.3 +/- 0.3 to 4.5 +/- 0.4 ng/ml. It is concluded that combined glibenclamide and NPH insulin at bedtime is useful to treat secondary failure in non-insulin dependent diabetic patients, but their response in variable and non dependent on their beta insular secretion.