SUMMARY We have identified regular thermal patterns over normal knee, ankle, and elbow joints and demonstrate how synovitis affecting these joints may be identified by alteration or loss of the thermal pattern. Sixty healthy volunteers were thermographed on a total of 190 occasions, and 614 out of 618 joints conformed to the normal thermal pattern. Eight-five patients with synovitis of at least one of the specified joints were thermographed on a total of 339 occasions, and 322 out of 1362 thermograms were abnormal. No joint with clinical evidence of synovitis had a normal thermal pattern. As temperature-based parameters have been found to show marked diumal variation and relative frequency distributions do not have this drawback, we suggest that quantification of synovitis by thermography should in future be based on abnormalities of thermal pattern rather than absolute skin temperature values.The objective assessment of disease activity in rheumatoid arthritis is difficult. Many parameters are based on patients' symptoms, which may not give an accurate indication of the progress of the disease, and laboratory evaluation can be unhelpful.' Infrared thermography (IRT) has been used for the past 10 years to measure skin temperature over inflamed joints.23 Horvath and Hollander4 measured the intra-articular temperature in patients with rheumatoid arthritis and noted that it could be used as a guide to the acuteness of inflammation. Bacon et al. 5showed that measurement of mean skin temperature could be used as a measure of disease activity. Skin temperature is affected by many internal and external factors.6 Circadian rhythm, metabolic rate, calorific intake, physical activity, emotional state, and atmospheric temperature and humidity can all exert a marked effect on skin temperature. In particular, owing to the proximity of blood vessels to the skin in the extremities even small alterations in vascular tone will produce large changes in skin temperature.Many of these factors are uncontrollable. In addition, the variation leads to difficulties in standardisation and relating a single measurement to an arbitrary 'normal' temperature.7We therefore set out to examine whether such relation to absolute arbitrary standards is necessary and whether inspection of the thermogram could identify the presence or absence of synovitis. Thermograms of the knee, elbow, and ankle joints of normal subjects have been compared with those from patients with clinical evidence of synovitis. To test the effect of diurnal variation on the methods of analysis serial readings were obtained at different times of day. Methods and patientsThermography. This was carried out in a draught-free room with ambient temperature controlled to 20 5°C + 0 5°C and humidity 50% + 10%. Subjects were seated in a modified dentist's chair with their limbs exposed and a sleeveless jacket worn.An AGA Thermovision 680 medical system was used to detect infrared emission, and all thermograms were recorded in digital form on magnetic tape for subsequent analysis with...