A multiple baseline design across behaviors was used to assess the effects of sound production treatment (SPT) with a speaker who had chronic apraxia of speech and aphasia. The speaker had participated in a previous investigation involving SPT and had demonstrated clinically significant levels of overgeneralization of trained sounds. In the present investigation, treatment was modified to attempt to reduce unwanted overgeneralization. Three groups of sounds were trained sequentially in the word-initial position: Group 1: s, p, v; Group 2: k, sh, j, ; and Group 3: l, m, n. Treatment consisted of a combination of repetition, production of minimally contrastive words, integral stimulation, and phonetic placement cues, with treatment stimuli presented in a combination of blocked and randomized trials. Treatment effects were measured to trained words, to trained sounds in the word-final position of untrained words, and to a sentence completion context. Results indicated positive acquisition effects, limited generalization across word position and elicitation contexts, and minimal overgeneralization.