The literature contains many measurements of the fatigue properties of compact bone, but these experimental results have been difficult to interpret and use due to a large amount of apparent scatter:​ variation in the number of cycles to failure for a given cyclic stress or strain range. Recently Taylor (1998a, Journal of Orthopaedic Research, 16, 163–169) showed that much of this scatter could be explained using a statistical model which took into account specimen size, or more specifically stressed volume. The present paper describes an attempt to test this model by using it to predict some new data, for bovine bone tested in compressive loading at room temperature at physiological loading rates. Twenty specimens were tested at the same applied load range (100 MPa). The theory was able to predict the mean behaviour of the specimens very well, with an accuracy (expressed in terms of stress) of 2%. It was also able to predict the degree of scatter (i.e. the variation of N_f), which was shown to be similar to that measured by other workers.