The paper is concerned with the physical and metallurgical changes that occur in hot forging dies. These changes are examined by the execution of simple upsetting tests on a Petro-Forge high speed hammer. Each test involves the forging of 1000 steel (En 8) billets using flat dies in two materials, No.5 and H.50 steels. The temperature conditions for each test are varied by changes in the dwell time, from 5 ms to 1 s, and by the use of either unlubricated dies or dies lubricated with a colloidal graphite-water mix. Macroscopic examinations of the surfaces, and microscopic examinations and microhardness tests of surface sections are carried out on each pair of dies. It is found that dwell time affects the condition of the die surfaces in two distinct areas, for both die materials. In the die centre, with lubricated dies, a mild form of cracking present at the shortest dwell time, disappears as the dwell time increases. In contrast, away from the die centre, where metal flow and die surface temperatures are likely to be at a maximum, severe cracking is present at the longest dwell time and this reduces to a minimum as the dwell time is reduced. There was little evidence of cracking in any area with unlubricated dies. The micro-examinations reveal the presence of a layer of martensite away from the die centre with lubricated dies, the layer thickness increasing with dwell time. It is also thicker with No.5 than with H.50 dies. No transformation is evident at the die centre, nor when unlubricated dies are used. It is argued that during hot forging there are four possible temperature-cycles per blow with respect to the minimum and maximum surface temperatures in relation to the austenite and martensite transformation temperatures, and in only one case is martensite found in the die surface. This case corresponds to the conditions reported, when a water-based lubricant is applied and when the dwell time is sufficiently long for austenite to be formed. The effects of the dwell time and of a change in die material prove this supposition. The effects of temperature cycling and phase transformations on the thermal fatigue of the dies is discussed. It is concluded that transformation is the major cause of the severe cracking created in certain die areas of the reported results. It is stated that the use of coolants, as opposed to lubricants, in hot forging is a compromise and can be detrimental except where long dwell or contact times lead to severe softening of the working surfaces of the dies.