Data concerning anticipatory coarticulation in speakers with an apraxia of speech are contradictory. Some researchers report delayed or deficient coarticulation in apraxic patients while others report preserved coarticulation. Acoustic, perceptual and electropalatographic (EPG) analyses were used to examine anticipatory coarticulation in one apraxic patient and one normal control. Subjects produced three repetitions of six words containing the vowels /i/, /a/ and /u/ in the phrase 'Say --- again' at slow, habitual, and fast speaking rates. The speech samples were 'gated', leaving 0, 1/4 or 1/2 of the vowel. Listeners chose the vowel they heard from a set of six (/a/,/ae/, /i/, /I/, /u/, /?/. As the vowel /a/ was identified correctly, and had minimal EPG contacts, the data are not reported. The results for vowels /i/ and /u/ are as follows: (1) listeners had greater difficulty correctly identifying the vowels produced by the apraxic patient; (2) for both speakers, /i/ was identified with greater accuracy than was /u/; (3) with respect to speaking rate, accuracy of vowel identification was lowest at fast speaking rates and highest at slow rates; (4) the initial consonant influenced listeners' judgements of the upcoming vowel; (5) EPG patterns of the apraxic patient were variable and often inappropriate for the vowel; (6) dialect of the speakers may have influenced the accuracy of identification of the vowel /u/. From acoustic analysis it was found that the speakers in this study produced the vowel /u/ as a diphthong. Overall, the data suggested that the apraxic patient's coarticulatory gestures were delayed or distorted at all rates of speaking. Clinical and theoretical implications of these results will be discussed.