Purpose: This investigation was designed to further the development of a treatment for acquired apraxia of speech (AOS), Sound Production Treatment (SPT), by examining the effects of blocked and random practice. Method: A multiple-baseline design across participants and behaviors was used with 6 speakers with chronic AOS and aphasia. Accuracy of production of target sounds in treated and untreated words produced in probe sessions served as the primary dependent variable. Stimulus generalization was also measured to phrase production and sentence completion. Participants received SPT applied with blocked presentation of treatment words (SPT-blocked) and SPT applied with random presentation of treatment words (SPT-random). Results: Increases in accuracy of articulation of target sounds in treated words were observed for all participants for both conditions of treatment. SPT-random appeared to be associated with better maintenance for 2 participants. Generalization to untreated words was positive for all participants for SPT-random and SPT-blocked. Stimulus generalization effects varied across participants and measurement conditions; patterns of generalization did not appear to be associated with treatment condition. Conclusions: There may be an advantage for SPT-random for some speakers with AOS. Findings from the nonspeech motor learning literature may not translate directly to the treatment of AOS.