We tested the hypotheses that nonselective β-adrenergic blockade does not cause absolute hypoglycemia unawareness but shifts the glycemic thresholds for symptoms to lower plasma glucose concentrations and that neither neuroglycopenic symptoms nor cognitive impairments during hypoglycemia are altered by β-adrenergic blockade. To do so, we applied the euglycemic and stepped hypoglycemic clamp techniques to patients with moderately controlled insulin-dependent diabetes mellitus (IDDM) in the absence (n = 8) and presence (n = 9) of the nonselective β-adrenergic antagonist propranolol. Compared with the corresponding euglycemic clamps, total symptom scores first increased at the 4.4-mM plasma glucose step (a higher level than that of 2.8 mM in nondiabetic subjects studied previously) in the absence of propranolol. β-Adrenergic blockade did not produce absolute hypoglycemia unawareness. Indeed, at the frankly hypoglycemic step of 2.8 mM, total symptom scores tended to be higher in the presence than in the absence of propranolol. This was largely the result of greater (P < 0.01) perception of diaphoresis. However, symptom scores did not increase until the 3.3-mM plasma glucose step during β-adrenergic blockade. The perception of hunger, and perhaps that of tremulousness, was reduced by propranolol at the higher glucose steps. Neuroglycopenic symptoms were not reduced by propranolol. The cognitive function of memory, but not that of attention, was impaired, also starting at the 4.4-mM glucose step. This was not impaired further by propranolol. Thus, we formed the following conclusions. 1) Nonselective β-adrenergic blockade does not cause absolute hypoglycemia unawareness but shifts the glycemic thresholds for symptoms to lower plasma glucose concentrations in patients with IDDM. 2) β-Adrenergic blockade does not reduce neuroglycopenic symptoms, and it does not further impair cognitive function during hypoglycemia in IDDM patients.