Acoustic Doppler current profiler (ADCP) velocity measurements an subject to bias due to the effect of the signal processing filters on the spectrum of the Doppler-shifted signal and on the noise. Bias will occur when the filter is not centered on the signal. Numerical models of the received signal and the processing filter used in RD Instruments profilers show that biases on the order of 10 cm s−1 can occur in the lower half of the current profile in regions of high current shear. Errors tend to increase with the width of the acoustic beam and with the speed of the ship, and decrease with the pulse length. These biases are identified in ADCP velocity measurements made in the high shear of the equatorial undercurrent. We suggest criteria for editing existing ADCP data to remove excessive bias, and we recommend changes in profiler parameters which should greatly reduce the bias in future datasets. Abstract Acoustic Doppler current profiler (ADCP) velocity measurements an subject to bias due to the effect of the signal processing filters on the spectrum of the Doppler-shifted signal and on the noise. Bias will occur when the filter is not centered on the signal. Numerical models of the received signal and the processing filter used in RD Instruments profilers show that biases on the order of 10 cm s−1 can occur in the lower half of the current profile in regions of high current shear. Errors tend to increase with the width of the acoustic beam and with the speed of the ship, and decrease with the pulse length. These biases are identified in ADCP velocity measurements made in the high shear of the equatorial undercurrent. We suggest criteria for editing existing ADCP data to remove excessive bias, and we recommend changes in profiler parameters which should greatly reduce the bias in future datasets.