Abstract Rates of abrasion by a blade scraper are compared and contrasted with those obtained using an abrasive surface (a DIN abrader) for six rubber compounds: carbon black filled and unfilled vulcanizates of NR, SBR and cis-polybutadiene (BR). Results from the DIN abrader were found to be less sensitive to the amount of frictional work, and less discriminating between different compounds, than those obtained with a blade scraper. These differences are attributed to somewhat different mechanics of tearing in the two cases, even when the same fracture criterion is employed. For a blade, the length of the contact line is unchanged as the applied load increases, whereas for an asperity an increase in load causes an increase in length of the contact area, tending to mitigate the effect of the increased load. Approximate relations are developed along these lines to compare the tearing action of many sharp asperities to that of a blade. The results account reasonably well for the observed differences between rates of abrasion in the two cases. It is also pointed out that the rate of abrasion is not a unique function of the amount of frictional work expended.