The bias and precision errors of digital particle image velocimetry are quantified. Uniform displacement images are used to evaluate the uncertainty attributed to various sub-pixel peak finding algorithms. Bias errors are found to exist for all algorithms, and the presence of bias error tends to affect the precision error. The ability to "calibrate" out the bias error is explored using a rectangular free jet experiment. The calibration was effective in removing the bias error in the potential core and less effective in the shear layer. The bias error is found to functionally depend on the displacement gradients present in the interrogation region. The study stresses the need for in situ quantification of DPIV uncertainty.