Attention has been drawn to the fact that during exposure of nickel to air at elevated temperatures, oxygen diffuses through the grain boundary and reacts with carbon forming gas bubbles of carbon monoxide and/or carbon dioxide. Voids are produced due to the high pressure generated by the gas bubbles which cause creep embrittlement. Alloying nickel with 3.8 weight per cent aluminium, which is a stronger oxide forming element as compared to carbon, could not prevent diffusion of oxygen during exposure of the alloy to air at elevated temperatures. The diffusing oxygen reacts even in the Ni-Al-C system, with carbon and forms gas bubbles of CO and CO/sub 2/ which lead to creep embrittlement. The presence of 15 weight per cent chromium in nickel (Inconel 600 and Inconel alloy X-750), however, plays an interesting role on the kinetics of oxygen diffusion. At temperature 1050 degree Celsius and below, during exposure to air at atmospheric pressure, oxygen can diffuse along the grain boundary and reacts with carbon forming gas bubbles of carbon monoxide, where as at 1120 degree Centigrade and above, the formation of 'barrier' oxide scale, chromium oxide, at the surface prevents diffusion of oxygen. On the other hand, the presence of chromium, as observed in Inconel alloy X-750, does not play an effective role in preventing diffusion of oxygen along the grain boundary in poor vacuum. Circumstantial evidence suggests that diffusing oxygen reacts with carbon and forms gas bubbles of CO with consequential creep embrittlement, reduction in creep life and enhancement in creep rate. It has been shown that exposure of the alloy in poor vacuum has a much more damaging effect on the creep properties than that of an atmospheric pressure. It is, therefore, emphasized that exposure of nickel base superalloys to poor vacuum at elevated temperatures must be avoided; in particular this effect is more pronounced in the alloy with thin section size. It is also shown that even in the absence of oxygen interaction prior to creep testing, section size has considerable influence on the creep properties which is believed to be caused by oxygen interaction during creep deformation. For the same section size, the creep properties may be different in specimens of different geometrices. But for the same ratio of cross-sectional area to perimeter (A/P), rupture lifetime becomes independent of specimen geometry. Failure analysis of an aero engine blade has been presented as a case study to emphasize that while designing a component, utmost care must be excercised to avoid thin sections.