The design of pressure components for very high levels of structural reliability involves taking account of three types of variations or uncertainties. These include (1) the normal range of materials properties measured under laboratory conditions; (2) the normal or calculated ranges of load and duty cycles; (3) the added range of variation in properties due to fabrication, environment, and aging, and the possible added range of load or duty cycles due to infrequent transient conditions. Direct testing and analysis can reasonably attain confidence of failure rates of 10 −3/yr. The assurance of still lower rates - to 10 −6/yr, desirable for nuclear applications, benefits from a combination of testing and analysis, but augmented by intensive use of actual field experience of failures or degradation found by inspection. The scopes and uses of two major data systems are described. These are (1) the Significant Events Evaluation and Information Network (SEE-IN), and (2) the Nuclear Plant Reliability Data System (NPRDS). Together these systems cover the cumulative experience of ∼1,000 unit-years of nuclear plant operation. Uses of these systems to refine operations, maintenance, rework, and in some cases design or specifications, is described.