There has been little research on word recognition skills of phonologically disordered children, although recent work has shown they tend to perform more poorly than age peers on phoneme-identification tasks using synthetic speech stimuli. Six phonologically disordered preschool-aged children and six typically developing controls were asked to identify CVC words in which either the final consonant or the medial vowel had been removed in steps. The stimuli were digitized natural productions of familiar words. For the silent-center task, the phonologically disordered children were significantly less accurate than peers for the two conditions with the least acoustic information. For the gating task, the phonologically disordered children were significantly less accurate than peers for the two conditions with the most acoustic information. They were less accurate than peers at identifying digitized words with no acoustic information deleted, although they were not less accurate with live voice. This suggests that perceptual representation of final consonants for the phonologically disordered children is so fragile that even such a small degradation as a reduction in bandwidth affects their performance. Divergent patterns of group differences for the two tasks are probably related to differences in the nature of acoustic cues for medial vowels and final consonants.