During the life of a temperature-sensitive resistor (t.s.r.), two processes are at work producing permanent changes in body resistance. One gives an increase in resistance, and is thought to be crack propagation; with constant current loading, it can result in failure by fracture. The other process gives a decrease in resistance, and is thought to be continued reaction of the constituent oxides; with constant current loading, it can result in resistance stabilization. With normal current loadings, the approach to failure with life of an unprotected rod can proceed at a rate such that the ‘cold’ resistance is nearly proportional to the square root of lifetime. Operational lifetime of a rod is defined as the time required for the ‘cold’ resistance to increase by 15% with constant current loading. The corresponding current rating decreases logarithmically with the required operational lifetime. On this basis, the current rating of a particular rod should be reduced by at least 50%. This need for a reduction in rating is in accordance with the results of conventional life tests.There seems to be little justification for the popular assumption that the rod form of t.s.r. is intrinsically a much inferior component to the bead. This view is an unfortunate consequence of the relatively very different rating standards adopted for the two items. Continuous operation of a typical rod at a current twice the rated value reduces its life by a factor of 20 to about 100h. On the other hand, continuous operation of a bead at the original rating gives a life of more than 10000 h. Suitable derating of the rod would eliminate the apparent difference between the two forms.