Semantic Feature Analysis (SFA) is a treatment technique designed to improve lexical retrieval by increasing the level of activation within a semantic network. The purpose of this study was to replicate the Boyle and Coelho (1995) study in which SFA was applied with a mild non-fluent aphasic individual resulting in improved confrontation naming of trained and untrained items but no generalization to connected speech. The present study investigated whether a comparable treatment effect could be demonstrated, and to what extent severity and type aphasia might impact overall outcome. SFA was applied to an individual with a moderate fluent aphasia secondary to a closed head injury. Gains in confrontation naming of both trained and untrained stimulus pictures were noted as well as measures of connected speech. Potential explanations for these findings are discussed.