Scatter in mechanical property data of a metallic alloy derives from two sources: a variable microstructure or an inadequate and sometimes badly executed test method. A quantified example of each source of scatter is given in this paper by examining two important high-temperature properties: uniaxial tensile creep and low-cycle fatigue. Creep-lifetime data of engineering alloys often show scatter-bands much greater than can be accounted for on the basis of well-understood physics of testing-induced scatter. Bounds to this material-induced scatter have been computed for a low-alloy ferritic steel dataset using a physically based creep model incorporating damage state variables. High-temperature low-cycle fatigue demands a more complicated test procedure than does steady-load creep and the large interlaboratory scatter found in recent round robin data from 26 laboratories in Europe and Japan has highlighted inadequacies in the standardized test method. A simple material model and fracture criterion has been used, in conjunction with a previously introduced testpiece-bending model, to predict testing-induced interlaboratory scatter for the two nickel-base superalloys reported upon in the round robin exercise.