Wave or vehicular action on an ice sheet as well as structural vibrations and thermally induced loading cause cyclic loading on an ice sheet. To better understand the effects of cyclic loading on the strength of sea ice, cyclic loading tests have been conducted at the University Centre in Svalbard (UNIS). In addition, the corresponding ice behaviour was modelled based on Cole (1995), thereby describing the viscoelastic response of saline ice subjected to a cyclic stress. The aim of this study was to design and execute a reproducible experimental campaign for saline ice subjected to a cyclic compression, and to model the stress-strain relationship of the ice. Specifically, the effects of frequency and displacement amplitude were studied. The laboratory-grown saline ice was frozen from a mixture of sea water and fresh water with a salinity of approximately 8. The structure was classified as S2 columnar ice through inspection of thin sections. The specimens were retrieved from horizontal and vertical cylindrical cores. The porosities of the specimens ranged from 22 to 34 ppt and the salinities from 2 to 4. The experiments were performed by applying a sinusoidal varying uniaxial displacement of one piston of the loading frame using a stepper engine. An initial compressive load (equivalent to 1 MPa) was reached by applying a constant strain rate. The model uses kinematics to describe the ice behaviour, which is explained by the line defaults in the ice lattice, so-called dislocations. The input parameters of the model are the central relaxation time of dislocation relaxation, the dislocation density and an empirically derived distribution factor. The model results were given by the amplitude and phase lag of the steady-state stress response for an applied sinusoidal strain. The experimental campaign proved to be reproducible and demonstrated the stress response of saline ice subjected to a cyclic compression well. However, some improvements of the experiments are recommended; most importantly, a higher resolution of the strain sensors and more stringent displacement control, such that the input strain can be defined. The results from the experiments furthermore showed a dependence of the energy dissipation on the loading frequencies, as well as a considerable influence of stress relaxation on specifically the first cycle of the tests. To compare the tests and the model, the strain signal was filtered to remove the influence of the strain sensor location. The parameter used to compare the experimental results to the model was the loss compliance, which describes the energy dissipation per load cycle and is derived from the area per loop of the stress-strain curve. A discernible trend was an apparent increase in the loss compliance for an increasing frequency or per consecutive test, which may be caused by an increase of the dislocation density. In conclusion, the experimental method provides a successful experimental campaign that demonstrates the energy dissipation per cycle. The model provides solid results for the steady-state response of saline ice subjected to cyclic compression.