Quantitative determination of insulin secretion is of importance both clinically and in research. The optimal method has not been established, although several different methods have been used. We determined the reproducibility of islet function parameters obtained by the glucose-dependent arginine stimulation test, and also studied the priming effect of arginine on subsequent acute insulin responses. The test measures the acute insulin (AIR) and glucagon (AGR) responses to i. v. arginine (5 g injected over 45 s) at fasting glucose and glucose concentrations clamped at 14 and above 25 mmol/l, as well as the glucose potentiation of insulin secretion (slope_AIR) and the glucose inhibition of glucagon secretion (slope_AGR). When the test was performed twice in seven healthy women (mean ± SD age 58.7 ± 0.5 years, BMI 27.6 ± 5.5 kg/m²), the AIRs to arginine had a within-subject coefficient of variation (CV) of 18.6 % at fasting glucose, 18.7 % at 14 mmol/l glucose and 16.3 % at above 25 mmol/l glucose. The CVs for AGR were 11.6, 14.9 and 8.9 %, respectively. The CV of the slope_AIR was 24 % and of the slope_AGR 17.2 %. The arginine priming study was performed in six healthy women (age 63.7 ± 0.3 years, BMI 28.0 ± 6.9 kg/m²). Saline or arginine (5 g) was injected at fasting glucose, followed by arginine (5 g) at 14 mmol/l glucose. There was no difference between the acute insulin or glucagon responses to arginine at 14 mmol/l glucose in the two conditions, suggesting that there is no priming effect of arginine on the subsequent acute insulin or glucagon responses. Therefore, this method is a good tool to determine insulin secretion as, apart from its good reproducibility, it also provides several important parameters of islet function. [Diabetologia (1998) 41: 772–777]