This paper presents a methodology for quantifying the effect of harsh environmental conditions on the reliability of human actions in performing complex physical operations. A review of current human reliability techniques confirms that there is a lack of methodology for quantifying human errors while conducting complex physical operations in extreme environments. The developed methodology is based on a hierarchical Bayesian network accounting for causal dependencies among environmental factors, human error modes, and scenario-based activities. Also, a new model is developed with three reference points (awareness of the situation, access to a system, and action) that derives human error modes (HEMs) from physiological failure mechanisms and helps an analyst identify the root causes of human errors. The proposed methodology is applied to estimate the likelihood of human error in two different scenarios in harsh operating conditions in floating offshore structures. The two scenarios are a set of different human activities in a workplace under defined operational and environmental conditions. The proposed methodology helps enhance the safety of human performance while considering effective physical factors. It will also help to reform current regulations for working in harsh environments.