In the issue of September 1986, Chang et al.1 presented an interesting article entitled "Norepinephrine Removal and Release in the Forearm of Healthy Subjects." These authors concluded that antecubital venous plasma norepinephrine is likely to be derived mainly from local release and not from the arterial plasma norepinephrine input. In order to confirm this statement in nine patients with heart failure, we used previously described methods 2 to measure plasma norepinephrine, epinephrine, and dopamine withdrawn simultaneously (by one physician and two nurses) from pulmonary artery (PA), radial artery (RA), and antecubital vein (AV) of the same arm. Norepinephrine values from AV were systematically lower (981 ± 310 vs 1115 ± 313 pg/ml) than those from RA (12% on average) and from PA (981 ± 310 vs 1051 ± 340 pg/ml), but these differences were not significant, and the correlation coefficient between these measurements was very good (r = 0.96 and r = 0.99, respectively). Epinephrine values (AV, 365 ± 782 pg/ml; RA, 376 ±801 pg/ml; PA, 431 ±939 pg/ml) were also lower in AV, but not significantly, and the correlation coefficient was also very good (r = 0.99). Dopamine values were higher in PA, but not significantly (AV, 96 ± 7 1 pg/ml; RA, 96 ± 69 pg/ml; PA, 110 ± 9 2 pg/ml), and the correlation coefficient was 0.96.We conclude that values of AV samples collected to determine plasma norepinephrine, epinephrine, and dopamine in heart failure patients are lower than those for either RA or PA samples. However, we believe that this easy way of collecting samples remains an acceptable method for determining circulating plasma catecholamines in this setting.