This study aims at the development of non-contact dry eye diagnosis based on an infrared thermal imager system, which was used to measure the cooling of the ocular surface temperature of normal and dry eye patients. A total of 108 subjects were measured, including 26 normal and 82 dry eye patients. We have observed that the dry eye patients have a fast cooling of the ocular surface temperature than the normal control group. We have developed a simplified algorithm for calculating the temperature decay constant of the ocular surface for discriminating between normal and dry eye. This study shows the diagnostic of dry eye syndrome by the infrared thermal imager system has reached a sensitivity of 79.3%, a specificity of 75%, and the area under the ROC curve 0.841. The infrared thermal imager system has a great potential to be developed for dry eye screening with the advantages of non-contact, fast, and convenient implementation.