Introduction Clinical practice guidelines for preservation of upper extremity recommend minimizing wheelchair propulsion forces. Our ability to make quantitative recommendations about the effects of wheelchair configuration changes is limited by system-level tests to measure rolling resistance (RR). We developed a method that directly measures caster and propulsion wheel RR at a component-level. The study purpose is to assess accuracy and consistency of component-level estimates of system-level RR. Methods The RR of N = 144 simulated unique wheelchair-user systems were estimated using our novel component-level method and compared to system-level RR measured by treadmill drag tests, representing combinations of caster types/diameters, rear wheel types/diameters, loads, and front-rear load distributions. Accuracy was assessed by Bland-Altman limits of agreement (LOA) and consistency by intraclass correlation (ICC). Results Overall ICC was 0.94, 95% CI [0.91–0.95]. Component-level estimates were systematically lower than system-level (−1.1 N), with LOA +/−1.3 N. RR force differences between methods were constant over the range of test conditions. Conclusion Component-level estimates of wheelchair-user system RR are accurate and consistent when compared to a system-level test method, evidenced by small absolute LOA and high ICC. Combined with a prior study on precision, this study helps to establish validity for this RR test method.