Marine structures are typically sensitive to the direction of wind and waves, especially in extreme metocean conditions. The extreme metocean conditions and their associated predicted directions are not easily reachable from traditional design methodologies. In this research, the most probable combinations of different extreme metocean conditions along with their associated direction are predicted for the HyWind Scotland wind farm, Scotland. To achieve this, the Hierarchical Bayesian Modeling approach is applied to define the Joint Probability Distribution Function (JPDF) of four combinations of metocean parameters, including wave direction, wind direction and wind-wave misalignment. The data is provided by the ERA-Interim dataset in 40 years (1979–2018). The JPDFs are composed of a marginal PDF of directional variables (a mixture of von-Mises Fischer distributions) and two conditional JPDFs which are defined to satisfy the periodicity and positivity of distribution parameters. Then, applying the Inverse First-Order Reliability Method (IFORM) to the JPDFs, the Environmental Contours (ECs) for four sets of metocean data are developed. The results show that extreme values obtained from ECs, including directional variables, are higher than the values of traditional linear ECs. The maximum 50-year extreme value of wind speed from the JPDF of wind direction, wind speed and wave height is 2 m/s higher than the same extreme extracted from the JPDF of wind speed, wave height and period. Another important observed point is that the direction at which the extreme of metocean parameters occurs is quite different from their dominant direction of wind rose or the most probable direction of their probability density function. According to the results, it seems for direction-dependent structures; the application of this method may lead to a more realistic presentation of joint occurrence of linear and directional metocean parameters.