The prediction of fatigue life for metallic components subjected to complex multiaxial stress states is a challenging aspect in design. Equivalent-stress approaches often work reasonably well for uniaxial and proportional load paths; however, the analysis of nonproportional load paths brings forth complexities, such as the identification of cycles, definition of mean stresses, and phase shifts, that the equivalent-stress approaches have difficulties in modeling. Shear-stress based critical-plane approaches, which consider the orientation of the plane on which the crack is assumed to nucleate, have shown better success in correlating experimental results for a broader variety of load paths than equivalent-stress models. However, while the interpretation of the ancillary stress terms in a critical-plane parameter is generally straightforward within proportional loadings, there is often ambiguity in the definition when the loading is nonproportional. In this study, a thorough examination of the variables responsible for crack nucleation is presented in the context of the critical-plane methodology. Uniaxial and multiaxial fatigue data from Ti–6Al–4V and three other alloys, namely, Rene’104, Rene’88DT, and Direct Age 718, are used as the basis for the evaluation. The experimental fatigue data include axial, torsional, proportional, and a variety of nonproportional tension/torsion load paths. Specific attention is given to the effects of torsional mean stresses, the definition of the critical plane, and the interpretation of normal stress terms on the critical plane within nonproportional load paths. A new modification to a critical-plane parameter is presented, which provides a good correlation of experimental fatigue data.