Multi-walled carbon nanotube (MWCNTs) and graphene nanoplatelet (GNPs) filled high-density polyethylene (HDPE) composites with randomly dispersed (DCN) and segregated (SCN) conductive network structures were fabricated by a solution-assisted mixing method. The damage self-sensing behavior of the resulting composites was investigated via in situ electrical-mechanical measurements. The results show that nanofiller type and conductive network structure significantly influence the damage self-sensing behavior of the composites. The relative resistance change (RRC) of HDPE/MWCNT composites during tensile deformation can be divided into three stages. Compared to HDPE/MWCNT composites with DCN structures, more robust conductive networks are formed in SCN structures, resulting in smaller RRC. The self-damage sensing behavior of all HDPE/GNP composites follows a similar trend, starting with a quasi-linear increase in RRC followed by a sudden rise induced by brittle fracture of the material. Nanofiller content was also found to affect the damage self-sensing behavior of the composites with a higher nanofiller loading corresponding to a lower damage sensing sensitivity. A modeling study based on tunneling theory was also conducted to further analyze the mechanism. In addition, the tensile properties of the composites were measured. This study provides some important information for development of smart structural materials.