Studies suggest that the recognition of depression by primary care physicians (PCPs) is most likely in more symptomatic and impaired patients. As part of a randomized effectiveness study in primary care patients with panic disorder, we examined the baseline characteristics of study patients who were recruited by waiting room screen procedure (n=69) versus patients who were referred to the study by their PCP (n=41). Patients referred by their physicians had a significantly higher frequency of panic attacks, more intense attacks, and more anticipatory anxiety on the Panic Disorder Severity Scale, while screen-identified patients were more medically ill and had worse physical functioning on the SP36. There were no differences in anxiety sensitivity, phobic avoidance, depression, other SF36 measures, disability, or medical service utilization. In conclusion, differences in referred versus screened patients are mostly specific for panic attack-related symptoms, consistent with the notion that patients with more prominent physical symptoms (i.e., panic attacks) are more often recognized and referred in busy clinical settings. The better medical status and physical functioning of referred patients may indicate greater physician recognition of panic in patients who appear less medically ill. However, the many clinical and functional similarities between these two patient samples suggests that symptomatic primary care patients with panic may not always be identified by their PCPs and argues for the value of population-based screening for panic in primary care.