The effects of some test condition variables on tire rolling resistance results are presented and discussed from the viewpoint of establishing good testing practice. A designed experiment study was made on simultaneous effects of changes in load, speed, and inflation pressure on a radial and bias tire. All main and interaction effects were significant except for speed-load interaction on the radial tire and load-load interaction (curvature effect) on each tire. Quantitative effects of unit changes in each variable at any level of the other variables showed that load control is the most critical. Rolling resistance was shown to be significantly higher on a Safety Walk test wheel surface than on bare steel. No significant difference in scatter of results was found, however. Rolling resistance of a set of light truck tires was esentially the same on worn as on new Safety Walk. Results on sets of replicate tires suggested that four tires in each set are enough to characterize the set. A warm-up time of 20-25 minutes was shown to be sufficient to develop equilibrium air pressure. Data are presented to show how a safe upper limit to the test speed for a given tire can be determined, and how a safe service speed could be estimated without requiring that the tire be run to failure.