Abstract The paper describes friction experiments on smooth icy surfaces at various temperatures. Speed and temperature variations for NR gum rubber shows that in principle the WLF transformation can be applied to friction on ice, and friction—temperature curves may be interpreted as being part of the friction master curve on smooth solid surfaces. Very near the melting point the friction decreases rapidly for all polymers, and this appears to be associated with changes in the nature of the icy surface. At low temperatures, the friction—temperature curves do not only agree in shape with curves obtained on glass but reach similar values in magnitude also. Black and oil extension effects are also similar to those expected from friction experiments on other surfaces. The blending of polymers, either in the gum state or filled with carbon black, broadens the region of high friction, enveloping the friction curves of the two polymers making up the blend. Depending on the temperature region, either one of the constituents of the blend dominates the friction behavior. It is instructive to speculate on the relevance of these experiments to winter tire performance. It is obvious that the very clean polished smooth ice track obtained only under great difficulty even in the laboratory is not realizable in practice, so that the ice on which tires operate is in every case contaminated with either snow, water, or dirt. In this case the coefficient of friction drops drastically, as is demonstrated in Figure 11. Nevertheless it is easily shown that compound effects do exist in tire skid measurements on ice. It is generally observed that NR-based compounds have a higher skid resistance on ice than SBR-based ones. This ranking is also observed in all experiments described in this paper and is explainable because of the position of the friction master curve. Even if practical speeds, as they occur in tire usage, are considered, it is easily demonstrable that the shift of the master curve towards lower temperatures is insufficient to have a real impact on compound relations. It is in any case minimized by the fact that, at high sliding speeds, the simultaneously occurring temperature rise in the interface tends to cancel any velocity shift in the master curve. On ice the temperature rise cannot exceed 0°C, so that only a very small band of the master curve is in operation. In this band, NR is near its maximum, SBR has exceeded its maximum value, and BR has not reached it.