The objective was to investigate the reliability and effects of age and noise on high-frequency hearing thresholds. A cross-sectional study was used involving 187 exposed and 52 non-industrial noise-exposed subjects selected randomly from noise-exposed and non-industrial noise-exposed subjects, respectively. Each subject was tested with both conventional-frequency (0.25-8 kHz) and high-frequency (10-18 kHz) audiometry. Test-retest results showed that high-frequency audiometry (HFA) was as reliable as the conventional procedure. Although the inter-subject variation was large, the intra-subject variation was small, indicating that HFA can be used more reliably than the conventional procedure to monitor individual cases over time. Both the hearing threshold at high frequencies and the upper frequency limit deteriorated as a function of age and frequency. The exposed subjects had significantly higher hearing thresholds than the non-exposed subjects at all the high frequencies tested, the difference between the two groups being greatest at 14 kHz. Multivariate analysis indicated that age was the primary predictor and noise exposure the secondary predictor of hearing thresholds in a high frequency range (10-18 kHz). In contrast, multivariate analysis indicated the reverse order-noise exposure as the primary predictor, then age-for a conventional frequency range (0.25-8 kHz). The results of this study suggest that HFA might be used as an early indicator for noise-induced hearing loss and acoustic trauma rather than audiometry at a conventional frequency (4 kHz), particularly for younger groups.