The contribution describes the effect of a neonatal hearing screening program in terms of estimated prevalence rate of congenital hearing impairment and age at identification in two five-year cohorts born between 1990 and 1994 as a function of health authority districts (HADs). In addition, identically defined five-year birth cohorts from 1970 to 1974 and 1980 to 1984 living in the same HADs evaluated previously are used in the analysis, offering longitudinal data. In 1990 a non-targetted neonatal hearing screening program based on EOAE was introduced in the County-HAD, whereas the City-HAD continued its child hearing health surveillance program unchanged. Assuming an unchanged prevalence estimate of 1.5 per 1000 of congenital or early acquired (i.e. neonatal period) hearing disability, i.e. > or = 25 dB HL for the better ear at 0.5-4 kHz in both HADs, an underestimate of 68 per cent in the City and of 20 per cent in the County, respectively, was found at the time of data collection (January 1995). Significant longitudinal improvements in the early identification from the 1970-1974 cohort compared to the 1980-1984 cohort has been demonstrated, but the improvement in the proportion of children identified as a function of both 6 and 12 months of age in the 1990-1994 cohort is significantly greater, showing a median age at identification of 11 months in the County-HAD. The cross-sectional and longitudinal comparisons between the HADs imply that a neonatal universal hearing screening program with a 20 per cent coverage may add significantly to the early identification of children with early-onset hearing impairment.