To discriminate between true secondary failure (TF) and pseudo-secondary failure (PF) to oral hypoglycaemic agents, we studied 34 non-obese non-insulin-dependent diabetic patients who were being treated with these drugs. Nine were in good control (GC) with oral treatment, while 25 showed apparent SF. During a controlled hospital diet, fasting blood glucose remained persistently high in 15 of these patients (TF), while in the other 10 patients it clearly improved (PF). Fasting plasma glucose (FPG) and HbA1c were higher and body mass index (BMI) was lower in TF patients than in PF patients (P less than 0.01). C-peptide concentrations differed significantly among the three groups both in the fasting state (TF 0.25 +/- 0.02 nmol/l, PF 0.70 +/- 0.03 nmol/l, GC 0.74 +/- 0.03 nmol/l; P less than 0.0001) and 6 min after glucagon injection (TF 0.50 +/- 0.04 nmol/l, PF 1.02 +/- 0.06 nmol/l, GC 1.14 +/- 0.07 nmol/l; P less than 0.0001). C-peptide and plasma insulin curves obtained after a standard mixed meal also showed significant differences (P less than 0.001). In particular, there was a statistically significant difference between GC and PF versus TF (P less than 0.05), while there was no statistical difference between PF and GC. We conclude that some patients with apparent SF can improve their metabolic control if they strictly adhere to a correct diet (PF); a single measurement of basal C-peptide concentration or examination of the C-peptide and insulin responses to a meal are useful indicators for distinguishing patients with PF from those with TF to oral hypoglycaemic agents.(ABSTRACT TRUNCATED AT 250 WORDS)