Purpose <p>Frequency selectivity plays a crucial role in auditory perception, yet its precise characterization in humans remains debated. Most behavioral or physiologic estimates of frequency selectivity in humans have historically been obtained from individuals with clinically normal audiograms. However, emerging evidence suggests that even within this population, subclinical cochlear deficits may be prevalent, potentially skewing prior tuning estimates toward broader bandwidths. Here, we tested the hypothesis that human cochlear tuning is sharper when subclinical deficits, specifically hearing sensitivity above 8000&#xa0;Hz, are considered.</p> Methods <p>Using stimulus frequency otoacoustic emission (SFOAE) delays, we obtained physiological estimates of cochlear tuning sharpness (<i>Q</i>_ERB). We applied the Stockwell transform, optimized through <i>in silico</i> experiments, to accurately extract SFOAE delays in adults with normal audiograms (<i>n</i> = 37) while systematically accounting for hearing thresholds above 8000&#xa0;Hz.</p> Results <p>Our findings demonstrate that controlling for subclinical deficits results in significantly sharper (higher) tuning estimates at 2000 and 2828&#xa0;Hz among the tested frequencies. Additionally, applying a normative criterion for extended high-frequency hearing, we observed narrower equivalent rectangular bandwidths—approximately two times sharper—in individuals with better extended high-frequency sensitivity. The <i>Q</i>_ERB estimates aligned closely with forward masking data in the literature, reinforcing the view that humans possess sharper cochlear tuning than common laboratory animals.</p> Conclusion <p>These findings highlight the influence of subclinical hearing deficits on cochlear tuning estimates and suggest that humans with optimal cochlear health may possess even sharper biological frequency selectivity than previously reported.</p>