Two methods are commonly used to estimate the critical shear stress of individual fractions in mixed-size sediment, one using the largest grain displaced, the other using the shear stress that produces a small value of transport rate for each fraction. The initial-motion results produced by the two methods are typically different: largest-grain critical shear stresses vary with roughly the square root of grain size, and reference transport critical shear stresses show little variation with grain size. Comparison of the two methods is seldom possible because both methods can rarely be applied to the same data. The one case known for which both methods can be used suggests that the typical differences in initial-motion results reflect more methodological influence than real differences in the initial motion of different sediments. Although the two classes of methods may not be directly compared, a general definition of initial-motion in mixed-size sediment is presented that allows the characteristic differences between the results to be explained in terms of sampling and scaling considerations inherent in the mixed-size initial-motion problem. The initial-motion criterion defined also provides a rational basis for collecting comparable and reproducible data using the two classes of method.