The ignition of flammable atmospheres from hot surfaces arising from mechanical interactions has been a significant cause of many industrial and mining explosions. An investigation of the surface temperature generation resulting from sliding friction during short duration mechanical impacts has been carried out and the nature and usefulness of dynamic thermocouple measurement examined in the context of predicting mechanical ignition. The experimental results reveal that there is only a limited relationship between the measured maximum temperatures and the tangential energy loss during an impact. This appears to be mostly due to variation of the extent to which the tangential energy loss represents frictional loss (associated with tip sliding) rather than material deformation. Whilst an increase in impact energy tends to raise the measured surface temperature, there is significant random variation under nominally similar conditions. It is considered that this is associated with the randomness and changing nature of the contacting areas. During the small time-period of a mechanical impact, there is insufficient time for any equalisation of temperature between neighbouring contact zones to take place. With reference to the ignition of flammable gases brought about by mechanical impact, surface temperatures measured by dynamic thermocouple appear to offer only limited predictive usefulness since they could be associated with contact areas of insufficient size to transfer enough energy into the gas mixture to cause ignition. Finger-marking impact surfaces has the effect of greatly reducing the frictional energy loss but this is not fully reflected in the measured maximum surface temperature. It is concluded that ignition prediction should still be based on tests conducted with mechanical impacts taking place in an ambient flammable atmosphere.