This study examined the possibility that children who omit word-final stops as a clinical entity may preserve the voicing contrast of those omitted stops by differential durations of the receding vowel. Data bearing on this hypothesis were obtained by constructing a speech sample in which the voicing of word-final stops was systematically manipulated and by having three children produce these utterances under controlled conditions. Acoustic measures of vowel durations showed that for all word-final stops in both isolated words and words in sentence-final position that were judged unanimously by three transcribers to be omitted, two of the children produced vowel durations that were clearly sensitive to the voicing characteristic of the (omitted) final stop. The third child failed to show a consistent difference in vowel duration dependent on the voicing of the final stop. A phonological analysis of spontaneous speech sample showed that the two children who produced the reliable vowel duration difference also showed clear phonetic evidence of medial stops when the form containing the omitted final stop was inflected. The third child, however, omitted stops in both inflected and noninflected forms. These results are discussed in terms of the conditions that may be necessary for a child to use differential vowel durations to mark the voicing of omitted, work-final stops.