OBJECTIVES: To examine the validity of several thermogram-derived indices of autonomic functioning in the diagnosis of reflex sympathetic dystrophy (RSD). DESIGN: A series of chronic pain patients were classified diagnostically based on thermogram results using discriminant function analysis, and validity measures (e.g., sensitivity, specificity) were used to determine the accuracy of computerized thermographic pixel analysis in discriminating RSD from other pathology. SETTING: The study was conducted at the Rush Pain Center, a multidisciplinary outpatient pain clinic. PATIENTS: A series of 46 chronic pain patients referred for suspected sympathetically mediated pain. INTERVENTIONS: All patients underwent computerized thermographic examination under a baseline condition after acclimating to a climate-controlled room, immediately after a cold challenge was applied to the contralateral uninvolved extremity (4 degrees C for 90 s) and 20 min after the cold challenge. OUTCOME MEASURES: Temperature during the three experimental periods, degree of temperature asymmetry between affected and nonaffected limbs during the three periods, response to cold challenge, and recovery following cold challenge were measured. RESULTS: Temperature asymmetry accurately discriminated between RSD and non-RSD patients, with the most accurate asymmetry measures obtained at baseline. Responses to cold challenge and actual temperature values did not discriminate between RSD and non-RSD pain patients. CONCLUSIONS: Thermography can be a useful component of RSD diagnosis. In situations where sensitivity and specificity are equally important, an asymmetry cutoff of 0.6 degree C appears optimal. If specificity (i.e., accurately ruling out non-RSD cases) is more important, a cutoff of 0.8 degree C or 1.0 degree C may be considered as well.