An 18-section computer model of the vocal tract was used to investigate articulatory-acoustic relations by mapping from an articulatory feature space to an acoustic formant frequency space. The articulatory space was defined in terms of a five dimensional matrix in which the dimensions represented (1) degree of raising of the front of the tongue, (2) degree of raising of the back of the tongue, (3) extent of lip aperture, (4) degree of rounding, and (5) height of the larynx. The tongue dimensions and the lip dimensions were rescaled to satisfy a definition of equal articulatory step size in terms of similar differences in the total midsaggital section of the tongue and in the square root of lip area. Examination of the formant frequency space suggested that changes in tongue shape alone produce unique sets of values for the first three formant frequencies. However, variations in tongue shape combined with variations in lip area and larynx height result in articulatory compensation giving equivalent values for the first three formants. Changes in tongue shape alone failed to generate plateaulike regions of acoustic stability. [Work supported by NIH.]