Several recent studies have demonstrated that good glycaemic control reduces the risk of late diabetic complications such as retinopathy, nephropathy and neuropathy [1±3]. However, maintenance of good glycaemic control considerably increases the incidence of all degrees of hypoglycaemia [4]. The magnitude of this increase has been shown to be closely related to the achieved level of glycaemic control [4]. In patients with a diabetes duration of more than 15 years, in patients with strict glycaemic control and in those who previously experienced a severe hypoglycaemia blunted counterregulatory hormonal responses [5,6] and absence of autonomic warning symptoms to hypoglycaemia have been reported [7]. Hypoglycaemia unawareness, i. e. the inability to perceive consciously the onset of hypoglycaemia, may, among other factors, be induced by even mild episodes of hypoglycaemia, including nocturnal hypoglycaemia [8±10]. In well-regulated insulin-dependent diabetic (IDDM) patients nocturnal hypoglycaemia occurs once every 2±3 nights [11] and this incidence rate might even be an underestimation [12]. Probably the high incidence of nocturnal hypoglycaemia is mainly a consequence of the characteristics of the Diabetologia (1998) Summary In patients with insulin-dependent diabetes mellitus (IDDM) good glycaemic control confers an enhanced risk of hypoglycaemia. Nocturnal hypoglycaemia occurs frequently and contributes to the syndrome of hypoglycaemia unawareness. In order to avoid nocturnal hypoglycaemia we substituted night-time continuous subcutaneous insulin infusion (CSII) therapy in 14 patients with well-controlled IDDM using a multiple injection regimen for the more variable bedtime NPH insulin. During a stepwise hypoglycaemic clamp we studied the effect of this regimen on counterregulatory hormonal responses, warning symptoms and cognitive function. In addition, we investigated the incidence of daytime hypoglycaemia and the acceptability of night-time CSII treatment. CSII was associated with a lower frequency of hypoglycaemia (mean ± SEM): 16.1 ± 3.1 vs 23.6 ± 3.3) episodes during the last 6 weeks of treatment, p = 0.03 (CSII vs NPH)) with maintenance of good glycaemic control (HbA 1c 7.2 ± 0.2 vs 7.1 ± 0.2 %, p = 0.2). Hypoglycaemic thresholds for the growth hormone response and for autonomic symptoms were lower for CSII treatment than for NPH treatment. Of 14 patients 6 decided to continue with the nocturnal CSII treatment. In conclusion, nocturnal CSII improves warning symptoms and counterregulatory hormonal responses to hypoglycaemia and is an acceptable treatment strategy for patients suffering from hypoglycaemia unawareness, as demonstrated in this acute feasibility study. [Diabetologia (1998)