Acoustic Doppler current profilers (ADCPs) are commonly used to measure streamflow and water velocities in rivers and streams. This paper presents laboratory, field, and numerical model evidence of errors in ADCP measurements caused by flow disturbance. A state-of-the-art three-dimensional computational fluid dynamic model is validated with and used to complement field and laboratory observations of flow disturbance and its effect on measured velocities. Results show that near the instrument, flow velocities measured by the ADCP are neither the undisturbed stream velocity nor the velocity of the flow field around the ADCP. The velocities measured by the ADCP are biased low due to the downward flow near the upstream face of the ADCP and upward recovering flow in the path of downstream transducer, which violate the flow homogeneity assumption used to transform beam velocities into Cartesian velocity components. The magnitude of the bias is dependent on the deployment configuration, the diameter of the instrument, and the approach velocity, and was observed to range from more than 25% at 5cm from the transducers to less than 1% at about 50cm from the transducers for the scenarios simulated. ?? 2007 ASCE.