Although uncertainties are present in all real parameters, numerical calculations of the structural behaviour of offshore wind turbines are usually conducted with deterministic values. However, with this approach, optimisation processes can be misleading and reliability levels cannot be calculated. The reasons for deterministic calculations are high computing times of probabilistic approaches and the lack of knowledge about the scatter of data. For deterministic approaches, more complex models with higher computing times are possible, although they, are less generally valid. Therefore, it is useful to identify the most influential parameters that have to be treated in a probabilistic manner using sensitivity analyses is valuable. Contrary to rudimentary sensitivity approaches being used in offshore wind energy so far, this paper presents a new four-step sensitivity analysis reducing the probabilistic parameter subset step by step and aiming to achieve a compromise between computing time and complexity. It can be shown that for different substructures and different load cases, only a small parameter subset is influential and many other inputs can be regarded as deterministic without losing accuracy. However, attention must be paid to the slight differences among substructures. Therefore, it must be highlighted that not all results are general. (C) 2017 Elsevier Ltd. All rights reserved.