Issues concerning corrosion service life predictions on the basis of laboratory measurements are examined. Models for corrosion service life prediction will reduce corrosion failures and operation costs by enabling better design, alloy selection, full life-cycle cost analysis, and operation. An expected value approach for estimating corrosion damage accumulation and failure rates is proposed. Examination of this approach indicates that understanding the uncertainties due to the variability of the exposure environments will be a critical element of these predictions. It is concluded that, with properly developed, evaluated, and standardized corrosion defect sample designs and propagation rate measurement techniques, it will be possible to develop corrosion defect tolerant design approaches analogous to those used successfully to prevent fatigue failures.