Safety-critical components, such as aircraft landing gear, are designed using the 'safelife' fatigue analysis process. Variability exists within materials data, loads data and component dimensions and is currently mitigated using safety factors. Probabilistic approaches to safe-life fatigue design have been proposed to better represent this variability. However, challenges currently exist that prevent the wider utilisation of a probabilistic approach. This paper presents a framework that aims to overcome these challenges. The statistical characterisation, probabilistic, surrogate modelling and sensitivity analysis methods required to implement the framework are introduced. Finally, a discussion of how recent advances within aerospace fatigue design, such as 'big-data', can be used to support a probabilistic framework is presented.