AbstractGrain yield stability is vital for achieving yield consistency across a broad range of environments. The significance of this is well documented in crop genetic studies and may equally be relevant for tillage practices used in croplands. The objective of this study was to evaluate long‐term winter wheat (Triticum aestivum L.) grain yield stability under different tillage practices. The study was designed as a randomized complete block with three replications. The independent variable was tillage with three levels: no tillage (NT), stubble mulch (SM), and moldboard plow (MP). The experiment was established as a winter wheat–fallow under a dryland cropping system. Each phase of the wheat–fallow was present each year. The historical grain yield data from 1972 to 2010 were presented in this study. Yield stability was assessed using regression coefficients (βi) and squared deviation from regression (S2d). Grain yield had average stability because each tillage practice had a βi of 1.0. The SM tillage had an S2d of 0.03 (P = .90), which was not significant than zero (0), while NT and MP had an S2d of 0.06 (P = .04) and 0.07 (P = .01), respectively, that were significant than zero (0). This suggests that SM had a more stable yield under different environments when compared with NT and MP. In general, using minimum tillage, such as SM, that maintains residues on the soil surface could contribute to yield resiliency across different environments and enhance land sustainability in dryland cropping systems.