ABSTRACT Current practice for protection against potential failure sources is presented. The variability in loadings and in the strength of structural elements typical for offshore installations is discussed in terms of statistical properties. On this basis a reliability analysis based on modern probabilistic methods is carried out. A comparison of the reliability of different types of structures is made depending on their structural elements, materials and design codes. INTRODUCTION The need for risk evaluation and reliability criteria is today gaining increased recognition in the offshore industry. Ideally such criteria should be quantitative and provide for unambiguous interpretation of safety targets and standards. Probabilistic reliability is the only meaningful concept which can be used to obtain a logical and objective distribution of risks and safety requirements in the offshore industry. Probabilistic reliability is a tool for specifying criteria for how safe offshore engineering works should be and for evaluating design and operational requirements from probabilistic criteria for the installation as a whole. The mathematical tool and the concepts used for the probabilistic calculations are defined in the appendix. FAILURE SOURCES Sources of failure of an offshore structure may be associated withdesignchoice or use of materialworkmanship and maintenanceoperation of the structure Failures are associated with accidental lack of compliance with - written or unwritten - criteria for safe design, construction or operation. This lack of compliance may be caused by errors which may conveniently be classified as suggested by Gauss 150 years ago:gross errors (blunders)systematic errorsrandom errors Examples of gross errors are a decimal point misplaced in the analysis, the use of low grade steel where high grade steel is specified etc. A common feature of all gross errors is that they can be eliminated by adequate care. On the other hand no design safety factor, however large, will give adequate protection against the effect of gross errors. For this reason the only way of eliminating gross errors is by extensive supervision and inspection of all activities involved in the process of design, construction, maintenance and operation of the structure. A reliable inspection must be based on the contractor's own supervision and control of his activities. In addition the quality of the control has to be inspected by some independent authority carrying out independent design evaluation and survey of the construction and operation. It must be realized that expenditure in time and money for these activities contributes to structural safety even more than large factors of safety. The systematic errors are associated with causes which contribute to deviations to one side. These errors may be associated f. inst. with empirical relationships between visual wave records compared to significant wave heights, recognized design formulae based on incomplete theory or experiment, strength properties determined by distorting recording devices. etc.