Seven consonant-vowel-consonant minimal pairs, differing only in the voicing characteristic of the final consonants, were elicited from 20 three- and four-year old native speakers of American English. Spectrographic analyses of the utterances revealed that (1) children produced vowel duration differences of the same nature and magnitude as those found in adult speakers' utterances; (2) the duration of a preceding vowel, as well as the duration of voicing during the final consonant closure, are reliable predictors of the voicing characteristic of the final consonant (3) other measures, such as syllable duration, final consonant duration, and vowel duration plus final consonant duration, are not as reliable as vowel duration and closure voicing duration as predictors of final consonant voicing; (4) the three- and four-year olds did not produce significantly different vowel or closure voicing duration.