In dye-sensitized solar cells (DSCs), the electron transfer from photoexcited dye molecules to semiconductor substrates remains a major bottleneck. Replacing TiO2 with ZnO is expected to enhance the efficiency of DSCs, owing to the latter possesses a much larger electron mobility, but similar bandgap and band positions as TiO2 remain. However, the record efficiency of ZnO-based DSCs is only 7% compared with 13% of TiO2-based DSCs due to the even slower electron-transfer rate in ZnO-based DSCs, which becomes a long-standing puzzle. Here, we computationally investigate the electron transfer from the dye molecule into ZnO and TiO2, respectively, by performing the first-principles calculations within the frame of the Marcus theory. The predicted electron-transfer rate in the TiO2-based DSC is about 1.15 × 10(9) s(-1), a factor of 15 faster than that of the ZnO-based DSC, which is in good agreement with experimental data. We find that the much larger density of states of the TiO2 compared with ZnO near the conduction band edge is the dominant factor, which is responsible for the faster electron-transfer rate in TiO2-based DSCs. These denser states provide additional efficient channels for the electron transfer. We also provide design principles to boost the efficiency of DSCs through surface engineering of high mobility photoanode semiconductors.