Purpose: The ultimate goal of aphasia therapy should be to achieve gains in function that generalize to untrained exemplars and daily conversation. Anomia is one of the most disabling features of aphasia. The predominantly lexical/semantic approaches used to treat anomia have low potential for generalization due to the orthogonality of semantic and phonologic representations; this has been borne out in a meta-analysis of treatment studies. The intensive, neurally distributed, phonologic therapy reported here can, in principle, generalize to untrained phonologic sequences because of extant regularities in phonologic sequence knowledge and should, in principle, generalize to production of words trained as well as those untrained. Method: Twenty-six persons with chronic aphasia due to stroke were treated, in a staggered (immediate vs. delayed treatment) open trial design, with 60 hr of intensive, multimodal therapy designed to enhance access to and efficiency of phonemes and phonologic sequences. Results: There was an absolute increase of 5% in confrontation naming of "untrained" nouns at 3 months, and there were 9% to 10% increases on measures of generalization of phonologic processes. Conclusion: The results of this trial demonstrate generalization of training effects on laboratory measures, which were sustained at 3 months, and provide support for the theories that motivated the treatment.