A proper understanding of abrasion resistance and associated damage mechanisms is of vital importance in the design of improved abrasion resistant construction steels. The conventional scratch test, sliding a rigid indenter under a controlled load and speed against a smooth surface, mimics the nature of the abrasion process and can be used to evaluate the abrasion resistance of various microstructures. However, scratch tests are mostly done on the initial surface, which can be very different from that formed during the abrasion process and hence do not truly reflect its abrasion resistant response. In the present work, a new scratch test methodology is developed to approach the real abrasion condition by carrying out a dual-pass dual indenter scratch tests, in with a small indenter scratches a pre-scratched surface produced by a large indenter. Five steel grades with different work hardening capacities, i.e. Interstitial-Free ferritic steel (IF steel), Fully Martensitic steel (FM steel), Dual Phase steel (DP steel), Quench Partitioning steel (Q&P steel) and TWining Induced Plasticity steel (TWIP steel) were selected. Systematic scratch resistance experiments were performed to investigate the damage mechanisms, the work hardening behavior and the development of subsurface deformation layers. Results suggest that the work hardening layer formed beneath the abraded surface plays the dominant role in determining the abrasion resistance. Steels grades of DP, Q&P and TWIP display superior scratch and abrasion resistances, notwithstanding their relative low hardness compared to that of a corresponding steel with a fully martensitic microstructure.