<p>This paper presents the first systematic numerical study employing a realistic elasto-plastic non-linear soil model to examine factors affecting the measurement of soil stiffness at low to intermediate strains in the triaxial apparatus. The study is motivated by the development of new image-based measurement techniques for triaxial testing. The results from numerical simulations are presented that quantify the influence on sample stiffness determination of three important features common to all triaxial tests, namely: the misalignment of the loading direction and the top of the sample (and indirectly bedding errors), the use of fixed or free load cell connections and the effect of system compliance. The analyses compare data obtained using four different instrumentation systems employed for axial strain measurement with overall predicted deformations and elemental response of a typical sand sample. The paper highlights some of the significant errors that can occur in standard triaxial tests and makes recommendations to assist more accurate determinations of sample stiffness. It is also shown numerically that, despite sample and test imperfections, the common approach of averaging data from at least two strain gauges located on a triaxial sample’s middle third can lead to reasonable inference of a soil’s elemental Youngs modulus.</p>