The aerosol mass measurement method, DMA-APM, measures a lower mass as compared to the electrical mobility diameter-based particle mass for sub-50 nm nanoparticles. The extent of underestimation increases with decreasing nanoparticle diameter and can reach as much as 20–80% for different nanoparticles between 10–20 nm. To study this issue, the DMA-APM system was tested with traceable size standards (PSL and NanoSilica) and laboratory generated silver nanoparticles. It was found that the extent of mass underestimation depended on Brownian diffusion as well as the strength of the classifying forces, and the extent was quantified by a dimensionless parameter λcP, which is suggested to be higher than 40 to eliminate the mass underestimation for size standards. Further analysis also showed that the uncertainty in the particle density of test nanoparticles should be as low as possible to minimize the error in the judgment on the accuracy of the APM. Finally, the absolute accuracy of the APM at different λcP, was determined by the size standards, which could be used to correct for the mass underestimation for sub-50 nm nanoparticles.