EDF Energy own and operate seven Advanced Gas cooled Reactor (AGR) nuclear power plants which are the only commercially operated high temperature nuclear reactors in the world. Being high temperature means that considerable numbers of components operate within the creep regime and as a result, creep-fatigue is a life limiting degradation mechanism for many reactor components. Nuclear safety is the overriding priority for the operator, EDF Energy. Therefore demonstrating the integrity of structural components is an important activity. This poses the greatest challenge for components within the reactor pressure vessel, such as boilers, insulation and support structures, because they can not be easily inspected or repaired due to accessibility. Therefore there is reliance upon theoretical structural integrity assessments to demonstrate components are safe to operate, using procedures such as the RS assessment procedure. This research reviews the requirements placed upon high temperature structural integrity assessments and the current solutions provided by the deterministic assessment procedures, using a systems engineering approach. This identified clear disparities, including: deterministic structural integrity output versus a probabilistic safety requirement; lack of communication about the extent and nature of uncertainty in deterministic assessment results; plant observations (survival, failure and inspections) cannot be reconciled with deterministic assessment results; deterministic assessments consider components in isolation and do not consider the functionality of a structural system. These disparities are of growing concern when considering the aged AGR plants, as component failures become more likely and lifetime extensions are sought. To address these disparities, a physics-of-failure reliability framework has been proposed as a solution for creep-fatigue assessments. An important aspect of this paradigm shift, from deterministic to probabilistic assessments, was in the management of uncertainty. Two classifications of uncertainty were defined to do this; aleatory uncertainty which is caused by variation within a structural system and epistemic uncertainty which arises from a lack of understanding about the structural system. These two definitions can be applied to all variables and uncertainties allowing them to be managed appropriately within the structural integrity reliability framework. Uncertainties which are involved in a structural integrity analysis, including material properties, operating conditions and geometric properties were explored, classified and quantified where possible. The use of plant data, including survival and failure data and inspection data was also considered within the reliability framework.