Cone penetration tests have been performed in two different calibration chambers (as well as rigid walled concrete bin), using penetrometers of two different sizes, in order to study the influence of size and boundary conditions on measured cone resistance. This is necessary for assessing the extent to which current calibration facilities are able to model natural bodies of sand. For the equipment used, diameter ratios ranged from 21.4 up to 48.3. Although cone resistance in loose sand (Diameter ratio = 15-30%) is virtually independent of boundary effects throughout this range, it is not so in dense sand (Diameter ratio = 90%). In the case, it would seem that a diameter ratio of about 50 is desirable for NC sand and probably about 100 for OC sand (at least for OCR 8, as used herein). For samples of less favourable size, the B1 and B3 boundary options cannot be relied upon to provide true upper and lower bounds in the normal sense. Typical penetration traces for cone and sleeve resistance are presented, and problems in their interpretation are discussed, It is shown that K sub 0 boundary restraint may prejudice the attainment of a cone resistance plateau, due to a build-up of lateral stress and it is shown how these stress changes are linked, with lateral volume changes and piston movements. For samples of inadequate size, it would seem that cone resistance may be substantially controlled by the lateral stress acting, and it is suggested that this may not be true of in situ behaviour.