The advent of technologic improvements in assessing the hearing of newborn infants has made possible the implementation of universal newborn hearing screening. Furthermore, selective screening based on high-risk criteria fails to detect half of all infants with congenital hearing loss. Although universal screening has been recommended by the National Institutes of Health and the Joint Committee on Infant Hearing Screening, data to support this recommendation have been incomplete, and the recommendation has been seen as without solid foundation by many in the pediatrics field. This study was designed to assess the feasibility, accuracy, and cost-effectiveness of a hospital-based hearing screening program for all newborns. Between 1992 and 1996, hospitals in Colorado with 100 or more births per year were targeted to participate in universal hearing screening of newborns. To date, 26 of 52 targeted hospitals, ranging in size from 40 to 3500 births per year, have implemented universal screening. A total of 41 796 infants were screened between 1992 and 1996. Screening was performed using automated auditory brainstem response, otoacoustic emission testing, or conventional auditory brainstem response, with follow-up testing performed on those infants who failed initial screening. Of 41 796 infants screened at birth, 2709 failed initial screening, and of 1296 who have completed reevaluation, 94 have been identified with congenital sensorineural hearing loss (75 bilateral) and an additional 32 identified with conductive hearing loss (14 bilateral). The frequency of bilateral congenital hearing loss requiring amplification therefore is shown to be at least 1 in every 500 newborns. During the study period, an additional 17 children with significant hearing loss not identified until >/=18 months of age were reported voluntarily; all 17 had been born at hospitals not participating in newborn hearing screening. The false-positive rate for the screening program to date in Colorado is calculated to be 6%, but evolving technology has resulted in improvements to as low as 2%. Positive predictive value of an abnormal screen result is shown to be at least 5%, and as high as 19%, with improving technology. The sensitivity of newborn screening is demonstrated to be at or near 100%. Costs of screening are compared with other screened congenital diseases; although the true cost per child for newborn hearing screening is significantly higher than screening tests performed on blood, the much higher incidence of congenital hearing loss results in a comparable cost per case diagnosed when compared with hypothyroidism or phenylketonuria, for example. The feasibility of early intervention is demonstrated, with amplification by the use of hearing aids being the catalyst for effective treatment. Finally, the costs of screening and early intervention are compared with the monetary savings in avoiding delayed and therefore intensive therapy and intervention for children not diagnosed at birth. The true cost of screening for one newborn is shown to be between $18 and $33, with an average cost of $25 per infant. The cost per case of congenital hearing loss diagnosed is approximately $9600. A model for cost predictions and subsequent intervention savings is presented, and recovery of all screening costs is demonstrated after only 10 years of universal screening in Colorado. Universal newborn hearing screening is feasible, beneficial, and justified, as indicated by the frequency of the disease, the accuracy of screening tests, the ability to provide early intervention, the improved outcomes attributable to early amplification, and the recovery of all screening costs in the prevention of future intervention costs. Furthermore, the incidence of bilateral congenital hearing loss is alarming, and is, in fact, many times greater than the combined incidence of all newborn screening tests currently performed on blood samples.