Extensive CFD modelling of cooling using nitrogen jets showed that an array of high velocity gas jets close to its surface could cool the part at a similar speed to oil. The optimum conditions were: an approximately uniform nozzle field with the jets four to eight times their own diameter apart, at a distance from the part to be quenched of a quarter of the diameter of the jets; and a jet velocity of 100 m/s. When these optimised conditions were applied to an idealised gear form, the model suggested that it could be fully hardened if a nitrogen/hydrogen mixture was used. The model was validated by comparison with physical experiments under exactly the same conditions. Unfortunately, although close to the physical results, the model results had some important differences. Part of the difference was explained by the exclusion of radiation losses from the model and part by the use of values for specific heat that were derived from static, rather than dynamic, experiments. When the model was modified to correct these there was closer, but by no means perfect, agreement. The finite element model used at this stage was thought to model the heat transfer accurately, but not the steel. The heat transfer data was therefore applied to a metallurgical model. The results from this model were superior in some respects, particularly regarding the phase changes occurring, although again not perfect, perhaps because of the experimental technique used.