Observable means of communicating pain and suffering, "pain behaviors," have been postulated to comprise an important construct relevant in both the development and the maintenance of chronic pain [Fordyce, W.E. (1976). Behavioral Methods for Chronic Pain and Illness, C.V. Mosby, St. Louis, Mo.]. Two groups of professionals who have direct contact with chronic-pain patients (i.e., physicians and psychologists) participated in a study designed (a) to identify the latent or underlying characteristics of pain behaviors and (b) to assess the degree of agreement of these characteristics between health professionals with very different training. Multi-dimensional scaling and hierarchical clustering statistical techniques were employed to identify the latent structure of pain behaviors. Two primary pain behavior dimensions were identified, namely, audible-visible and affective-behavioral. Four clusters of pain behaviors were identified and labeled distorted ambulation or posture, negative affect, facial/audible expressions of distress, and avoidance of activity. The two samples of health-care providers identified virtually equivalent latent characteristics of pain behaviors. The data suggest that there is consistency in the pain-behavior construct and that the latent structure is generally congruent with Fordyce's original conceptualization. The results provide an empirically derived basis for the assessment of pain behaviors.