Autogenous shrinkage is a complex phenomenon in which many physical and mechanical mechanisms are involved. The latter may also interact between each other and many of these coupling mechanisms are still not well understood. Understanding of these mechanisms and their quantification can provide tools for preventing early age cracking and its treatment. The aim of this thesis was to quantify in the first step the influence of the cement characteristics and the microstructural development on mechanisms occurring on a microscopic level leading to shrinkage and internal stress development. In the second step, a numerical model for determination of autogenous shrinkage was developed. A systematic experimental study has been performed, from the end of mixing up to 10 days, on series of cement pastes prepared from the same white cement with different finenesses and various water-to-cement ratios (W/C). Different characteristics were measured by a number of techniques in the experimental study: chemical shrinkage, hydration kinetics (isothermal calorimetry, XRD, SEM), relative humidity, capillary porosity (SEM, MIP), linear and volumetric autogenous shrinkage, elastic properties (ultrasonic propagation measurements). Using different techniques allowed to validate characteristics by more than one technique and to determine the action mechanism of each parameter. In order to provide an independent prediction, except for the intrinsic mechanical or physical properties of the existing phases which may determined by nano-indentation, the numerical model does not depend on any other experimental data from this study. The experimental and numerical results are compared and the differences are analysed.