Abstract Laboratory abrasion data were obtained on various tread vulcanizates over a large range of severities, including the region equivalent to the low severity encountered on straight, smooth California highways. The constant transmitted power principle was used as a severity criterion for comparison of abrasion of different tread vulcanizates. Load, wheel angular velocity, and ambient temperature were held about equal to those prevalent during road tests. These data were compared with road tests on the same vulcanizates, conducted on tires tested on courses varying from low to high severity. The laboratory data and the California low severity test data show a high degree of correlation for BR vs SBR (50 HAF) over the entire range of severities tested. The loss rate-severity plots for BR and SBR show a crossover point, with the BR showing the higher abrasion resistance at the higher severities. The BR shows a substantially higher abrasion resistance than the SBR on the Texas straight highway course. This is in opposition to the California data, which shows a substantially lower abrasion resistance of the BR compared to SBR on the straight highway course. The abrasion rates for BR and SBR in Texas, California, and laboratory tests are consistent if the severity of the Texas course is assumed to be equivalent to a linear combination of at least two California or laboratory severities. A plausible cause of the difference between California and Texas results is the difference in pavement surface texture. This viewpoint is shown to be consistent with abrasion patterns observed on tires in the California and Texas straight highway tests. The laboratory data and California test data on NR relative to SBR are consistent at the high severities, and for laboratory data extrapolated into the low severities from the high severity data. However, the actual laboratory data taken at low severities are more erratic and show consistently higher loss rates than those observed in the California freeway test. It is postulated that this may be the result of increased susceptibility to thermal oxidative degradation of the NR surface, caused by relatively longer exposure in the laboratory than on the road at a given severity. The investigation described in this paper lends a great deal of encouragement to the usefulness of laboratory abrasion testing for the screening of treadstock vulcanizates. The ability to obtain abrasion results under carefully controlled conditions of transmitted power, temperature, load and wheel angular velocity allows results to be investigated confidently as a function of conditions in the instrument, or of the polymeric properties. These investiagtions can be quite fruitful in eventual understanding of the polymer behavior or of the mechanism of abrasion. An important observation has also been the large variations in tire test results on different courses of assumed equal severities. Thus, the laboratory abrasion test can also assume the position of a guide to more effective road tests.