Suction caissons are used more and more for various in the oil&gas and offshore wind industries. Although, the use of suction caissons is not new, uncertainty still exists regarding their installation, due to the varying soil profiles encountered and the absence of experience within the offshore industry. Prediction methods, regarding the suction requirement, are not always seen to provide accurate estimations, but mostly provide a range of expected values. The general theoretical understanding of the different geotechnical issues arisen during suction caisson installation is known but not defined or quantified sufficiently. The main problem discussed in the thesis is the variation in soil characteristics encountered at offshore sites. Owning to this fact standardization of installation behavior and installation related parameters is challenging. There are a number of uncertainties in the installation prediction of suction caissons. First, the state of stress and soil conditions adjacent to a suction caisson being installed differ from those around typical driven piles or drilled shafts. Dynamic changes are imposed changing the soil state. Second, the soil resistance encountered during the installation of suction caissons depends on the rate of installation, hydraulic conductivity, drainage length, as well as the shear strength properties of the foundation soil material. Finally, during installation, volume characteristics of the surrounding soil change compared to those measured in-situ initially. The existing knowledge related to the prediction of soil resistance and installation of suction caissons is found to be adequately accurate in a relatively short range (homogeneous sand and clay profiles) of soil conditions. The grey area in between permeable and impermeable soils is found to be uncharted. The objective of the present research is to assess the governing mechanisms during installation of a suction caisson in layered sand by investigating the installation behavior and how prediction methods can be modified based on a back-analysis of executed installations. The limitations of existing methods are investigated regarding the soil resistance prediction. The accuracy level associated with the suction requirement in sand and layered sand is evaluated. The monitored installation pressure is assessed in order to verify consistent patterns of installation pressure trends. Typically, in this thesis, installations of suction caissons in homogeneous dense sand profiles have been observed to meet the theoretical predictions regarding the soil resistance encountered during installation. Estimation with adequate accuracy level of the associated suction requirement was observed. Conversely, the installations of suction caissons in layered sand with varying soil characteristics (permeability and relative density) are observed to be inadequately described by the prediction methods regarding the installation suction pressure requirement. Adjustment of predictions’ input parameters was seen to be required based on experience with the particular soil material, in order to return reliable estimations. Parameters such as prediction methods’ ? P?_su^crit were seen to depend on the encountered soil material and its characteristics, determining their loosening rate. Furthermore, the ? P?_su^crit was noticed to be essentially the parameter determining the anticipated soil plug loosening rate based on the analyzed prediction methods. A refinement of the adopted loosening rate by prediction methods for various soil profile characteristics (i.e. initial relative density and permeability) was seen to be required to enhance installation pressure estimation accuracy. The analysis of layered sand profiles interbedded by layers with fine-grained material were seen to behave as virtual sand profiles rather than layered sand profiles (silty sand profiles), when permeability was remained high. Regardless, of the soil profile encountered, the installation pressure was seen to be a function of both the CPT cone resistance integral and the corresponding effective vertical stress at the c depth. DNV standard recommendations have been seen to be conservative, and without adequate specifications on k_f and k_p values, which are essential as they relate the CPT cone resistance with the estimation of the friction and tip resistance. Furthermore, a recommended suction-assisted caisson installation phase expression was seen to be required to standardize the suction caisson installation design. Further studies on the variation of the DNV k_f and k_p values in regards to the various soil characteristics are required.