Photoacoustic (PA) tomography is a rapidly developing imaging modality thatprovides high-contrast, high spatial-resolution images for vessel distributionsin tissue. It can be applied to early breast cancer detection, and therefore itwill be a valuable method for breast cancer diagnosis. Tissue absorbs andscatters light, and the optical fluence is known to approximately decreaseexponentially. The pixels or voxels in a reconstructed PA image represent thelevel of absorbed optical energy, which is the product of the absorptioncoefficient and the optical fluence. Therefore, the contrast of tumors in deeptissue decreases because the optical fluence is low. Quantitative photoacousticimage reconstruction has been proposed to resolve this problem, but the processis based on compensating the reconstructed image with a pre-calculated opticalfluence distribution. Because the contrast-to-noise-ratio (CNR) in thereconstructed images of deep tissue is low, amplification also magnifies thenoise, which decreases the image quality. Here we propose a novel adaptive depthattenuation compensation algorithm that can provide greater imaging depthwithout degrading the CNR. The proposed method is evaluated by numericalsimulation and a phantom experiment. The results of simulation and the phantomexperiment indicate that the proposed method performs better than conventionalmethods.