This thesis describes an integrated approach to fatigue cracking, reliability and inspection of offshore structures. The basis of the approach is statistical in nature and draws on recent experimental data and field measurements. It is intended as a working tool for those engaged in design, structural appraisal and sub-sea inspection of steel jacket structures. A review of current practice has been made and the requirements of an integrated approach are established. An approach is proposed comprising a series of compatible models dealing with fatigue cracking, the reliability of cracked joints and the inspection of welds for fatigue cracks. The primary linking parameter is the distribution of fatigue crack size which is considered as a time dependent variable. An integral part of the approach is a new statistically-based fatigue crack growth model. This is developed and the parameters involved in the model estimated from an analysis of experiment and oceanographic data. For any fatigue calculation the model allows the corresponding fatigue crack growth distribution to be estimated for any time during, or beyond, the nominal fatigue life. A number of example calculations are included; and using one of these a Bayesian procedure for revising fatigue lives in the light of inspection results is demonstrated. The effect of fatigue cracking upon the various modes of tubular joint failure is considered using linear statistical models. Example calculations are performed for a typical joint. An inspection strategy is proposed based on the concept of minimising life costs, including risk costs arising from the consequences of possible structural failure. This allows alternative inspection plans to be evaluated and compared, and a typical example calculation is included. The approach is discussed in the context of possible alternative approaches and areas for further related research are identified.