The production of voice onset time (VOT) contrasts at two rates of speech was investigated in three groups of subjects: nonfluent and fluent aphasic patients and non-neurological controls. Subjects produced the consonants [bdgptk] in the environment preceding the vowels [i e a o u] in a carrier phrase at a slow/normal and a fast rate of speech. For the normal speakers, acoustic analyses revealed significantly shorter VOTs at the fast as compared to the slow/normal rate, with a larger change evident in the voiceless relative to the voiced stop consonants. Both groups of aphasic patients produced rate changes that were smaller in magnitude than those of the normal subjects. Further, both brain-damaged patient groups produced VOTs that were shorter in the fast condition compared to the slow condition. However, a number of aberrant patterns emerged. For the nonfluent aphasic subjects, as in previous studies, voiced and voiceless consonants were produced with somewhat overlapping VOT distributions, suggesting a deficit in speech timing in these subjects. For the fluent aphasic patients, no differences in magnitude of VOT change emerged for voiceless relative to voiced consonants. Implications of these findings for theories of the nature of speech production impairments in aphasic patients are considered.