Reliable evaluation of stress-strain characteristics can be done only in the laboratory where boundary conditions with respect to stress and strain can be controlled. The most popular laboratory equipment is a triaxial apparatus. Unfortunately, standard version of triaxial apparatus can reliable measure strain not smaller than 0.1 %. Such accuracy does not allow to determine stiffness referred to strain range most often mobilized in situ i.e. 10-3 ÷ 10-1%, in which stiffness distribution is highly nonlinear. In order to overcome this problem fundamental modifications of standard triaxial apparatus should be done. The first one concerns construction of the cell. The second refers to method of measurement of vertical and horizontal deformation of a specimen. The paper compares three versions of triaxial equipment i.e. standard cell, the modified one and the cell with system of internal measurement of deformation. The comparison was made with respect to capability of stiffness measurement in strain range relevant for typical geotechnical applications. Examples of some test results are given, which are to illustrate an universal potential of the laboratory triaxial apparatus with proximity transducers capable to trace stress-strain response of soil in a reliable way.