Modeling of random cyclic strain-life (CSL) relations of engineering material should be a basis of strain-based fatigue reliability analysis. A statistical model for the relations of a nuclear engineering material, 1Cr18Ni9Ti stainless steel pipe-weld metal under temperature of 240℃, is presented. In the model, a verified distribution, i.e. lognormal distribution, is used as an appropriate assumed distribution of the material fatigue life data. Based on the Coffin-Manson law, the relations are modeled by mean value-and standard deviation-cyclic curves of the logarithm of fatigue life. Then, fatigue analysis at an arbitrarily given probability can be made conveniently according to the normal distribution function. An approach for estimating the curves and their confidence bounds is developed by a linear regression technique. Different from the existent reliability analysis methods that considered the material constants in the law as independently random variables, present work treats them as dependently random variables from the fit of test data. Availability of the model has been indicated by an analysis of the material test data.