This paper reports on a longitudinal study of the acquisition of the voicing contrast in American-English work-initial stop consonants, as revealed through instrumental analysis of voice onset time characteristics. Four monolingual children were recorded at approximately two week intervals, beginning when the children were about 1;6. Data provide evidence for three general stages: (1) the child has no contrast and produces both adult voiced and voiceless stops with short lag voicing; (2) the child has a contrast but one that falls within the adult perceptual boundaries of one (usually voiced) phoneme, and thus is presumably not perceptible to adults; and (3) the child has a contrast that resembles the adult contrast. Three children had reached stage 3 by age 1;9 the fourth child was still in stage 2 by the age 2;4. In addition, the data document the development of the voicing contrast across the three places of articulation. The rate and nature of the developmental process are discussed briefly in relation to two competing models for phonological acquisition (the "across-the-board" model and the "lexical diffusion" model) and two hypotheses regarding the skills being learned. (Author)