The attenuation of excitation power reaching the focus is the main issue that limits the depth penetration of highresolution imaging of biological tissue. The attenuation is caused by a combination of tissue scattering and absorption. Theoretical model of the effective attenuation length for in vivo mouse brain imaging has been built based on the data of the absorption of water and blood and the Mie scattering of a tissue-like phantom. Such a theoretical model has been corroborated at a number of excitation wavelengths, such as 800 nm, 1300 nm , and 1700 nm ; however, the attenuation caused by absorption is negligible when compared to tissue scattering at all these wavelength windows. Here we performed in vivo three-photon imaging of Texas Red-stained vasculature in the same mouse brain with different excitation wavelengths, 1700 nm, 1550 nm, 1500 nm and 1450 nm. In particular, our studies include the wavelength regime where strong water absorption is present (i.e., 1450 nm), and the attenuation by water absorption is predicted to be the dominant contribution in the excitation attenuation. Based on the experimental results, we found that the effective attenuation length at 1450 nm is significantly shorter than those at 1700 nm and 1300 nm. Our results confirm that the theoretical model based on tissue scattering and water absorption is accurate in predicting the effective attenuation lengths for in vivo imaging. The optimum excitation wavelength windows for in vivo mouse brain imaging are at 1300 nm and 1700 nm.