In a study of the Bauschinger effect, data were collected from three sources in the published literature. Quantitative stress-strain data were taken from these papers, and the results re-analyzed. The resulting database has 44 lots of sheet steels, including drawing quality, interstitial free, bake hardening, HSLA (and related grades), dual phase, TRIP, recovery annealed, and martensitic grades. In analyzing the data, it is found that use of the 0.05% yield strength on reversal instead of the conventional 0.2% yield strength provides more generality in explaining the results. In this analysis, the Bauschinger effect is characterized by a term (BE), which is the difference between the steel strength just prior to reversal and the 0.05% yield strength on reversal normalized by the strength just prior to reversal. An initial prestrain of 2% is needed to establish a dislocation morphology that can be generalized across many of the steel grades. For steels with a predominantly ferrite microstructure and no added interstitial elements, BE exhibits a single monotonic trend line with respect to the strength just prior to reversal. For the bake hardenable grades, the 0.05% yield strength on reversal is slightly greater than the trend line. For the DP980, TRIP, and martensite grades, the 0.05% yield strength on reversal is considerably greater than the trend line. These steel grades either initially or after deformation have a greater volume fraction of martensite. The recovery annealed steels, which possess a much greater dislocation density, also exhibit 0.05% yield strengths that are significantly greater than the trend line. For steels with a predominantly ferrite microstructure over a wide range of strength, the single characteristic curve indicates that the initial reverse yield strength can be determined from the strength at the point just prior to the reversal. For modelers of sheet forming processes, the relationship between strength and yield on reversal can be used in material constitutive equations.