BACKGROUND: Glucagon-like peptide 1 (GLP-1) has been proposed as a treatment for type 2 diabetes. We have investigated the long-term effects of continuous administration of this peptide hormone in a 6-week pilot study. METHODS: 20 patients with type 2 diabetes were alternately assigned continuous subcutaneous infusion of GLP-1 (n=10) or saline (n=10) for 6 weeks. Before (week 0) and at weeks 1 and 6, they underwent beta-cell function tests (hyperglycaemic clamps), 8 h profiles of plasma glucose, insulin, C-peptide, glucagon, and free fatty acids, and appetite and side-effect ratings on 100 mm visual analogue scales; at weeks 0 and 6 they also underwent dexascanning, measurement of insulin sensitivity (hyperinsulinaemic euglycaemic clamps), haemoglobin A(1c), and fructosamine. The primary endpoints were haemoglobin A(1c) concentration, 8-h profile of glucose concentration in plasma, and beta-cell function (defined as the first-phase response to glucose and the maximum insulin secretory capacity of the cell). Analyses were per protocol. FINDINGS: One patient assigned saline was excluded because no veins were accessible. In the remaining nine patients in that group, no significant changes were observed except an increase in fructosamine concentration (p=0.0004). In the GLP-1 group, fasting and 8 h mean plasma glucose decreased by 4.3 mmol/L and 5.5 mmol/L (p<0.0001). Haemoglobin A(1c) decreased by 1.3% (p=0.003) and fructosamine fell to normal values (p=0.0002). Fasting and 8 h mean concentrations of free fatty acids decreased by 30% and 23% (p=0.0005 and 0.01, respectively). Gastric emptying was inhibited, bodyweight decreased by 1.9 kg, and appetite was reduced. Both insulin sensitivity and beta-cell function improved (p=0.003 and p=0.003, respectively). No important side-effects were seen. INTERPRETATION: GLP-1 could be a new treatment for type 2 diabetes, though further investigation of the long-term effects of GLP-1 is needed.