To improve the measurement accuracy and efficiency of void-underneath detection of concrete pavement, a novel identification method based on transmissibility function analysis was proposed. Numerical analyses were conducted to validate the method’s effectiveness. Firstly, a nine-slab 3-D finite element model of concrete pavement was developed in ABAQUS software, and then the acceleration responses induced by an impact load were collected for transmissibility function analysis; Secondly, the measurement point pairs were defined in two transmitting directions. Then, the two transmissibility-based identification indicator (TDI) matrices were calculated, representing the differences between the transmissibility functions of damaged status and undamaged status; Finally, 11 scenarios, including different types and degrees of void-underneath, were analyzed. Moreover, the effects of measurement points’ distribution and noise levels on identification performances were studied. The analysis results show that the two TDI matrices are suitable for the identifications of corner-void-underneath and edge-void-underneath. Both two TDI matrices are sensitive to corner-void-underneath, while only one TDI matrix whose transmitting direction is perpendicular to the corresponding edge is sensitive to edge-void-underneath. The effects of center-void-underneath on the two TDI matrices are not significant, as the distribution of void-underneath is not consistent with the distribution of TDIs. For multi-area void-underneath, the trends of the two TDI matrices are similar to the single-point void-underneath. The distribution of measurement points affect the identification performance significantly. Denser measurement points would provide more accurate identification results. In addition, this identification method is very sensitive to the noise, high noise level results in poor identification performance. It is recommended to use this method under low noise level.