Flux measurement by eddy correlation from aircraft is becoming increasingly useful for understanding the spatial pattern of air-surface exchange over heterogeneous surfaces. It is however, an order of magnitude more complex than that from fixed towers. Not only must the air motion be measured relative to the sensor, but also the sensors' motion relative to the earth, along with the flow distortion required to stay airborne. The “Best Aircraft Turbulence” (BAT) system addresses these requirements in a compact research-grade package of sensors, data acquisition, and software suitable for use on small airplanes. It is used by approximately 10 research groups worldwide. Modern Global Positioning and Inertial Navigation Systems (GPS/INS) provide highly accurate and well-characterized measure of the probe's motion with respect to the earth. Less well characterized is the measurement of the wind relative to the probe. The BAT probe is a pressure sensor with nine ports in a 130 mm hemispherical head which carries within 100 mm of each other sensors of both the air's motion and the probe's motion. A particular BAT probe received a thorough wind-tunnel test at Purdue University, but the results were difficult to generalize. A new test has been made by NOAA/ARL and Harvard University in the Wright Brothers Wind Tunnel at Massachusetts Institute of Technology. Although several non-generic features were found, acceptably generic patterns of systematic error were quantitatively characterized in the reported dynamic pressure and incidence angles as functions of the known speed and direction of the incident flow. Knowing these patterns allowed estimation of the contribution of the probe's systematic error to uncertainty in the vertical component of turbulent wind and hence to the uncertainty in fluxes. A demonstration of the resulting error will be presented for summer CO₂ flux in Illinois, at dawn before significant photosynthesis and in late morning with strong downward flux of CO₂.