||We present a statistical study of CCD (or CMOS) camera response to small images. Diffraction patterns simulating particle images of a size around 2–3 pixels were experimentally generated and characterized using three-point Gaussian peak fitting, currently used in particle image velocimetry (PIV) for accurate location estimation. Based on this peak-fitting technique, the bias and RMS error between locations of simulated and real images were accurately calculated by using a homemade program. The influence of the intensity variation of the simulated particle images on the response of the CCD camera was studied. The experimental results show that the accuracy of the position determination is very good and brings attention to superresolution PIV algorithms. Some tracks are proposed in the conclusion to enlarge and improve the study.