Nanofillers are widely used in polymeric matrixes for reinforcement primarily through the formation of 3D filler networks. The nanocomposite mechanical properties strongly correlate with the filler 3D network strength and quality as affected by the filler geometry, size and the dispersion quality. However, defining the network quality remains a challenging concept due to the difficulty of assessing nanofiller dispersion, the geometrical complexity of the 3D filler network as well as the unfeasibility of network direct observation. In this study, we compare two nanocomposite systems, one reinforced with carbon nanotubes (CNT) and the other with branched carbon nanotubes or carbon nanostructures (CNS). Thermoplastic polyurethane (TPU) nanocomposites were prepared via in-situ polymerization with the nanofillers dispersed in the polyol prior to the polymerization reaction. Microscopic observations of polyol suspensions and characterization of their rheological behavior were used to assess filler network quality. This study demonstrates that branched carbon nanotubes (CNS) form stronger filler networks by comparison with the straight CNTs, which translates into better mechanical performance of the corresponding nanocomposites. In addition, the presence of branches allows for rheological percolation threshold at much lower nanofiller content for CNS systems by comparison with the CNT’s. Computer simulations of nanofiller network formation by straight versus branched carbon nanotubes confirm our experimental observations.