OBJECTIVE Comparison of the effectiveness of combined therapy vs. an insulin regimen in NIDDM patients with secondary failure of oral hypoglycemic agents. RESEARCH DESIGN, PATIENTS AND METHODS: 27 NIDDM patients were randomly allocated to Group A (n = 14, insulin) and Group B (n = 13, insulin and sulfonylurea) with crossover after 3 months. After the next 3 months a decision was made about the further treatment according to the metabolic control. The patients were then treated for another year with the more successful regimen. Metabolic control, residual beta-cell secretory capacity, degree of peripheral insulin resistance (clamp) and insulin dose were followed during the whole study. RESULTS (median, interquartile range, in brackets; *, statistically significant difference at P < 0.05): the combined therapy was better than insulin alone in 2/3 of patients. Glycemic control was better (HbA1c at 3 months: Group A = 7.9(1.1)% vs. Group B = 7.0(0.5)%*; HbA1c at 6 months: Group A = 7.4(1.5)% vs. Group B = 8.1(1.5)%. Insulin dose was lower during the combined therapy in the first 3 months: Group A = 0.62(0.18) U/kg body weight vs. Group B = 0.39(0.16) U/kg body weight*. Combined treatment was associated with increased C-peptide excretion both fasting and postprandially. No significant differences in peripheral insulin resistance were noted between the two groups. The combined treatment remained successful even after one year. The two groups of patients with different effective treatment did not differ significantly in any of the observed parameters. CONCLUSIONS the combined therapy was more effective than insulin alone. Its favourable effect persisted after treatment for a year. It seems better to start the treatment of the oral failure with combined therapy compared with insulin first and later followed by combined therapy. On the basis of the observed parameters it is impossible to determine in advance which kind of treatment is more suitable for the individual patient.