We studied the correlation of plasma and urinary epinephrine (E) and norepinephrine (NE) levels with anxiety symptoms in three patient groups: 1) pheochromocytoma (PH+) (n = 17); 2) hypertensives with elevated catecholamine levels shown not to have a PH (PH-) (n = 25); and 3) patients with panic disorder (PD) (n = 23). Structured interviews and four self-rated anxiety scales were used: the SCL-90R Anxiety and Phobic Anxiety scales, and the Spielberger State/Trait Anxiety Inventories. The SCL-90R Somatization scale (which measures 12 somatic symptoms) was also utilized. None of the PH+ patients met DSM-III criteria for PD. Two met criteria for generalized anxiety disorder (GAD). Of the PH- patients, two had PD, two had GAD, and three had both. Urinary and plasma E did not show significant positive correlations with any of the four anxiety scales in any of the three patient groups. In both the PH+ and PH- groups, E was significantly correlated with the SCL-90R Somatization scale. NE was not significantly correlated with any of the four anxiety scales in the PH+ group. In contrast, in the PH- group, plasma NE was significantly correlated with anxiety on all anxiety scales (r = +0.55 to +0.77, p less than 0.05). Furthermore, in the PH- group, plasma NE was significantly correlated with those items of the SCL-90R Anxiety scale measuring the cognitive rather than the noncognitive symptoms of anxiety. In the PD group as well, plasma NE showed a significant correlation with the SCL-90R Anxiety Scale (r = +0.67, p less than 0.05). Taken together, our observations suggest that: 1) the effects of catecholamines in the periphery derived from a source independent of nervous system control (such as a PH) are not sufficient to elicit an anxiety disorder meeting DSM-III criteria; and 2) in patients without an autonomous source of peripheral catecholamines, NE in the periphery results from sympathetic nervous system activation and probably reflects, rather than causes, anxiety.