Existing scales to measure trust in physicians have differing content and limited testing. To improve on these measures, a detailed conceptual model was constructed and a large item pool (n = 78) was generated following a detailed conceptual model and expert review. After pilot testing, the best-performing items were validated with a random national sample (n = 959) and a regional sample of HMO members (n =1,199). Various psychometric tests produced a 10-item unidimensional scale consistent with most aspects of the conceptual model. Compared with previous scales, the Wake Forest physician trust scale has a somewhat improved combination of internal consistency, variability, and discriminability. The scale is more strongly correlated with satisfaction, desire to remain with a physician, willingness to recommend to friends, and not seeking second opinions; it is less correlated with insurer trust, membership in managed care, and choice of physician. Correlations are equivalent with lack of disputes, length of relationship, and number of visits [corrected].