Referee whistles have been suggested as a significant contributor to noise-induced hearing loss. Thirteen models of sport whistles were tested for sound power with a 3-m hemispherical array of 19 microphones. The whistler produced nine tweets of low, medium, and high effort with two samples of each whistle model. Sound power levels ranged between 74 and 115 dB re 1 pW. The low, medium, and high effort tweets had average power levels of 85 ± 6 dB, 100 ± 6 dB, and 110 ± 4 dB, respectively. Preliminary damage-risk analysis of the whistle impulses yield varied estimates for the allowable number of tweets before auditory damage might be expected. For the Auditory Hazard Assessment Algorithm, between 4 and 66 tweets may exceed the daily exposure threshold. Based upon the amount of eight-hour equivalent A-weighted energy, approximately 120 to 500 tweets would exceed the daily 85 dBA exposure limit. The directivity of the sound power measurements will also be examined, and risk of hearing loss will be discussed.Referee whistles have been suggested as a significant contributor to noise-induced hearing loss. Thirteen models of sport whistles were tested for sound power with a 3-m hemispherical array of 19 microphones. The whistler produced nine tweets of low, medium, and high effort with two samples of each whistle model. Sound power levels ranged between 74 and 115 dB re 1 pW. The low, medium, and high effort tweets had average power levels of 85 ± 6 dB, 100 ± 6 dB, and 110 ± 4 dB, respectively. Preliminary damage-risk analysis of the whistle impulses yield varied estimates for the allowable number of tweets before auditory damage might be expected. For the Auditory Hazard Assessment Algorithm, between 4 and 66 tweets may exceed the daily exposure threshold. Based upon the amount of eight-hour equivalent A-weighted energy, approximately 120 to 500 tweets would exceed the daily 85 dBA exposure limit. The directivity of the sound power measurements will also be examined, and risk of hearing loss will be discussed.