Marine renewable energy has the potential to form a key part of the renewable electricity generation arsenal that is required for society to prevent catastrophic global warming. Device reliability is one of the great engineering hurdles facing the sector; operating cost effectively in the harsh open ocean is very challenging. The nascent stage of the sector means that accurate reliability assessment of devices is difficult because of a lack of relevant data, uncertainty about loading conditions and a lack of consensus on device designs and architectures. This research develops and applies a suite of reliability prediction methods for use across the tidal turbine development life cycle. A standards based statistical approach is developed for the assessment of the entire turbine; the pitch system is found to have a particularly high failure rate. An empirical approach is developed and applied on a pitch system to allow for the relevant reliability influencing design parameters to be uncovered. A Bayesian framework is developed to enable new information to be incorporated into existing models; this is of particular interest to nascent technologies. Finally a Physics of Failure approach is developed for the pitch system roller bearing unit; measured flow data is used to predict the fatigue life and reliability via a hydrodynamic model. The methods presented along with the results will assist reliability assessment and prediction during tidal turbine design and early operation. Improved reliability assessment at each stage of turbine development can hopefully assist the industry to cross the ’valley of death’ and become commercially viable.