Several factors affecting the ultrasonic determination of stress in bolts are examined which help to clarify existing problems with the interpretation of certain experiments. To begin with, the use of ultra sonic waves to determine stress inverts the results of experiments used to evaluate third-order elastic constants. Thus, the unambiguous determination of stress must be subject to the same conditions as the experimental measurement of higher-order moduli. For example, Thurston and Brugger's expression for the transit time in terms of the natural velocity and the unstressed length provides an alternative to the usu. practice of relating the transit time to the stres·s dependent true velocity and the stress induced change in length. Using the natural velocity emphasizes the explicit stress dependence of the velocity and avoid the unnecessary corrections for changes in path length and density. Although temperature and dislocation mobility are closely controlled in third-order elastic constant determinations, these factors are rarely considered in ultrasonic stress measurements. It is shown that in steels a stress of ~10MPa (~1.5Ksi) is equivalent to a l°C change in temperature. The possible role of defects or temperature in the relaxation phenomena observed in ultrasonic stress determinations is also examined. The effects of constrained therm. stress and unloaded bolt length on the so-called stress-acoustic constant are presented. The correct functional form of the stress and strain dependence of the sound velocity is shown to be crucial to the problem of thermally modified bolt load. The effect of unloaded bolt length is considered for the cases where the nut is stationary during loading and where it is tightened to produce the load. Finally, the difficulties caused by longitudinal wave mode conversion upon reflection off the bolt sides are examined.