We show that the horizontal turbulent velocity components measured by the common sonic anemometer array can suffer attenuation and crosstalk as a result of the flow blockage caused by the acoustic transducer assemblies. Using an analytical model of this “transducer-shadow effect”, flow-blockage data from test arrays, and a simple linear model of the fluctuating response, we show the nature of the distortion in the measured velocity spectra. We suggest that rather than correct for the shadow effect, which ran be quite significant for horizontal velocity spectra and stress cospectra, it would be preferable to minimize it through design. There is encouraging evidence that the Kaijo-Denki transducer design produces much less shadow effect than the conventional (right circular cylinder) shape. Abstract We show that the horizontal turbulent velocity components measured by the common sonic anemometer array can suffer attenuation and crosstalk as a result of the flow blockage caused by the acoustic transducer assemblies. Using an analytical model of this “transducer-shadow effect”, flow-blockage data from test arrays, and a simple linear model of the fluctuating response, we show the nature of the distortion in the measured velocity spectra. We suggest that rather than correct for the shadow effect, which ran be quite significant for horizontal velocity spectra and stress cospectra, it would be preferable to minimize it through design. There is encouraging evidence that the Kaijo-Denki transducer design produces much less shadow effect than the conventional (right circular cylinder) shape.