A series of vowels similar to [ɪ] was synthesized using a different constant value of fundamental frequency (f0) for each member. As f0 ranged from 133 to 200 Hz, harmonics either fell exactly at the first formant frequency (F1) or a pair of harmonics straddled F1 with varying degrees of skew and thus differing relative amplitudes. The presumed task of the auditory system is to interpret harmonic amplitudes in the neighborhood of F1 in order to estimate the true F1 frequency of the vowel. It will first be shown that (1) energy-based methods of interpreting filter-bank outputs, (2) non-pitch-synchronous linear prediction analysis, and (3) synchrony measures based on auditory models all systematically mis-estimate F1 frequency by as much as plus/minus 10% as f0 varies. Secondly, a perceptual experiment was performed to determine whether these systematic estimation errors are present in auditory judgement data. Comparison of the standard stimulus series with one in which F1 is “corrected” so that a filter bank sees a constant F1 indicates that the uncorrected fixed-F1 series has more nearly constant phonetic quality. Accounting for the perceptual data in terms of simple processing strategies is not yet possible, but if the central auditory system is provided with the frequencies and amplitudes of harmonics near F1, a moderately complex calculation could yield F1 with minimal error. [Work supported in part by an NIH grant.]