Accurate measurement of cancellous bone’s apparent elastic modulus, E, is confounded by the experimental artefacts created when trabeculae are severed during specimen preparation. Although standardized axial testing protocols have been developed to deal with the so-called “end effects” caused by severed trabeculae at the loading surfaces, much less attention has been given to the “side effects” around the periphery and the specimen size dependence they create. Two models (one theoretical, one empirical) have been proposed in the literature to predict the reduction in E with decreasing specimen diameter. The current study used finite element method (FEM) modelling to analyze bovine cancellous bone from five different anatomic sites and quantify the changes in that occurred with specimen diameter. The two models were adapted so that they could predict E based on diameter and architectural parameters (BV/TV, DA, Tb.Sp) alone, without requiring that a “true” modulus be known a priori. Both models fit the data equally well;​ however, the empirical model gives simpler estimations as a function of trabecular separation (Tb.Sp). A minimum diameter of 5–8 Tb.Sp is recommended.