- To compare structured with intuitive methods of detecting exaggerated pain, in 2 randomized studies, with masked observers, and to compare properties of measures of relative tenderness, as surrogates for measures of pain. Pairs of experienced rheumatologists assessed tenderness and behavior in subjects who were either giving honest responses (Not Acting) or exaggerating their tenderness while hiding the deception (Acting). In Study 1 (20 subjects) the assessments of behavior were descriptive only; for Study 2 (25 subjects) a prestructured format was developed. The challenge to the examiners was to detect Acting. The challenge to the analyst was to develop, from the recorded data, an algorithm that described or bettered the examiners' performance. By exaggerating, nontender patients could meet formal ACR criteria for fibromyalgia. The addition of a prestructured pain behavior assessment in Study 2 was associated with improvement in the observers' ability to detect Acting, with improvement of sensitivity from 60 to 90% (p = 0.0003, Study 2 compared with Study 1). False positive diagnoses of Acting by the observers were uniformly uncommon, with specificity of 85 and 86% in the 2 studies. Formal numerical techniques are required and feasible for separately assessing the structural and psychogenic components of chronic pain.