For the development of Arctic offshore structures, design ice loads are required. These design loads represent the ice loads a structure may be exposed to for specified requirements. The design ice loads depend on magnitude of the ice loads and on the probability of exposure to the ice loads. Both are impacted by local sea ice conditions. Climate change affects the sea ice conditions causing change in the ice loads Arctic offshore structures are expected to experience. To obtain accurate design loads, the effect of climate change should be considered in defining those. Conventional methods to determine design ice loads are based on historical data and assume this data can be used to represent the loads during the lifetime of the structure. However, historical data cannot represent future ice conditions if climate change is considered as sea ice conditions will change. This means that the effect of climate change is not incorporated in the design ice loads when these are based on conventional methods. To include the effect of climate change a new method needs to be developed. In this thesis, such a new method is proposed that enables to include the effect of climate change into the design ice loads. Instead of historical data, the new method considers the future ice conditions. The method allows to base the design ice loads on sea ice conditions that change over time. To determine design ice loads, extremal distributions are used. Extremal distributions describe the probability of seasonal maximum ice loads. Commonly, the extremal distribution is based on data covering multiple seasons and cannot properly include inter-seasonal change in sea ice conditions. The new method allows to determine the design ice loads based upon changing extremal distributions. The extremal distributions are determined for seasons separately based on sea ice conditions of one season only. For the proposed method, a concept referred to as an 'ice state' is introduced. Ice states describe a period of time in which the sea ice conditions are assumed to be constant. When sea ice conditions are constant, the corresponding short-term ice load distributions and the ice load frequency can be determined. Both are required to be able to determine the extremal distribution for one season. According to the new method, the increase of drift speeds causes increase of design ice loads whereas the decrease of ice concentration, thickness and compressive strength causes decrease of the design ice loads. Based upon expected future climate change scenarios, the new method indicates that design ice loads are lower when compared to the design ice loads according to the conventional method.