Thermal efficiency of cold-stressed finger-tips during cold induced vasodilatation (CIVD) is considered. The actual heat loss from the finger-tip is compared to either the minimal or the maximal heat losses. The actual heat loss is estimated by integrating the area under the time-temperature curve of the finger-tip. The minimal heat loss is estimated by extrapolating an exponential approximation of finger-tip temperature until it reaches a certain minimal value. The value used in this study is 5 degrees C, which is the pain threshold. The maximal heat loss is calculated by assuming finger-tip temperature to be maintained at its initial value throughout the cold exposure. These quantities were calculated for a series of exposures involving two environmental conditions of gloved subjects: Tdry bulb = -17.2 degrees C, Tdew point = -25.1 degrees C (cold-dry) and Tdry bulb = 0 degree C, Tdew point = -8.4 degrees C (cold-wet). Thermal efficiency was in the range of 0.40-0.85 for the minimal heat loss value (eta min) and 0.22-0.72 for the maximal heat loss value (eta max). Weak linear relationships between the two definitions of the thermal efficiencies and the total duration of the CIVD phase was indicated. The thermal efficiency based on minimal heat loss indicated an inverse relation with the total duration of the CIVD phase. This contradiction could be reconciled by the application of the common concept of "coefficient of performance". Considerable inter- and intra-subjects variability was found.