A multi-segmental mathematical model of human thermoregulation was tested for its capability to predict individualized physiological responses. We compared the model predictions obtained for an average person with measured individual responses of subjects exposed to mild cold. Secondly, body composition (BC) data, the resting metabolic rate (MR), and the actual measured MR during the test were used as input into the model. The data was obtained from 20 subjects (age: 19–36 years; BMI: 17–32 kg/m2). BC, MR, rectal and skin temperatures were measured for 1 h at 22 °C, followed by 3 h at 15 °C. A mean bias of 1.8 °C, with a standard error of 0.7 °C, resulted for the mean skin temperature of an average person at 15 °C. When subjective BC and measured MR were incorporated the bias was -0.2±0.9 °C. For the hand-back skin temperature the bias ± standard error fell from 5.3±2.8 °C for an average person to 2.0±2.5 °C, when using individualized characteristics. Trunk skin temperatures were not significantly affected by the adjustments. In conclusion, this study shows that on a group level predictions of skin temperatures can be improved when adopting individualized body characteristics and measured MR, but the predictions on an individual level were not improved.