Photoacoustic tomography (PAT), which reconstructs the distribution of light-energy deposition in the tissue, is becoming an increasingly powerful imaging tool. For example, the technique has potential applications in the earlystage breast cancer sensing and the functional imaging of small animal brain. In PAT, the system signal-to-noise ratio (SNR) and the number of measurement positions (NMP) are the two main factors which affect the quality of final reconstructed image. Undoubtedly, the increase of SNR or the numbers of measurement positions will improves image quality. However, one has to pay a cost on the imaging speed for such improvement of image quality. In this paper, the factors influencing the imaging performance of PAT are investigated by means of computer simulations. The result shows that the increase of the number of averaging times in acquiring of acoustic signal and the number of measurement positions are efficient ways to improve image quality. However, there exists a turning point at which the further increase of NMP and averaging times makes the improvement of imaging performance negligible. Thus a tradeoff should be made to achieve the optimal reconstructed image according to the system SNR.