Tensile mean stress effects in notched metallic components are studied. Steel and aluminum stress-life fatigue data are analyzed using the generalized linear and Walker constant-life models with two classical approaches to mean stress adjustment, namely, the nominal mean stress method and the residual stress method. Particular attention is given to the number of free parameters when evaluating model performance. A generalization of the residual stress method is proposed both with one and zero additional free parameters. Analysis of the test data indicates that the Walker equation provides a better correlation for notched components than the generalized linear constant life relation and that the classical approaches to mean stress adjustment are overly optimistic (i.e., unsafe). The optimal degree of mean stress adjustment is shown to be sensitive to the relative stress gradient. The proposed zero parameter mean stress adjustment model improves model correlation performance in nearly all cases studied.