This investigation examined the phonologic similarity effect among individuals with aphasia (APH), with aphasia and apraxia of speech (APH/AOS) and normal controls (NC). Participants repeatedly produced pairs of rhyming words that contrasted with respect to the featural similarity of their onsets (i.e. shared voicing and manner, shared place and manner, no shared features). Rogers and Storkel (1998) used similar word pairs, but presented them one at a time, ostensibly requiring reprogramming of pre-motor processing buffers whenever a novel word was presented. They found that phonologic similarity delayed naming. In the present investigation, a parameter remapping task was used in which both words in the pair were presented before speakers began the rapid serial productions, thus alleviating the need to reprogram the processing buffer between the first and second word. Two measures of inter-word interval duration were obtained. The APH and NC groups showed no differences in inter-word interval length between the 'shared features' and 'no shared features' conditions but the APH/AOS group did. Since neither the APH nor the NC group exhibited the phonologic similarity effect in the context of a task that permitted two words to be programmed in advance, the susceptibility of the APH/AOS group to the featural similarity manipulation suggests that the capacity of their processing buffers may have been sufficiently reduced to have precluded programming of both words simultaneously. These findings support the hypothesis that one factor contributing to the disruption of normal speech production in AOS is a reduction in the capacity to program multiple syllables at a time.